From JoLandry at advaoptical.com Mon Aug 1 12:50:20 2016 From: JoLandry at advaoptical.com (Joan Landry) Date: Mon, 1 Aug 2016 12:50:20 +0000 Subject: [sysrepo-devel] verification options Message-ID: <89475774d3ee4888827638c2cbd47f39@advaoptical.com> When a user changes configuration objects to the candidate data store via a series of edit-config Netconf calls, after doing so issues a commit. Is there a mechanism provided by sysrepo to allow the device to reject the commit if the combination of changes would result in an invalid configuration of the device? Or does sysrepo just automatically commit the changes per the request? I was not able to find any info on this - would greatly appreciate any info you could provide on how the device can verify the datastore before it is allowed to be commited? Thanks, Joan Landry -------------- next part -------------- An HTML attachment was scrubbed... URL: From rastislav.szabo at pantheon.tech Mon Aug 1 12:54:15 2016 From: rastislav.szabo at pantheon.tech (=?iso-8859-1?Q?Rastislav_Szab=F3?=) Date: Mon, 1 Aug 2016 12:54:15 +0000 Subject: [sysrepo-devel] verification options In-Reply-To: <89475774d3ee4888827638c2cbd47f39@advaoptical.com> References: <89475774d3ee4888827638c2cbd47f39@advaoptical.com> Message-ID: <1470056055115.7797@pantheon.tech> Hi Joan, We do plan to support this, but it is not yet implemented. Currently sysrepo supports only post-commit notifications, meaning that callback is called AFTER the change has happened. Thi is signalled in the callback as event == SR_EV_NOTIFY. Any errors returned from the callback are ignored in this case. In the future, we would like to support callbacks being called even before the change has happened. This would require to specify SR_SUBSCR_VERIFIER flag in the subscription and it will be signalled as event == SR_EV_VERIFY in the callback. If that callback returns an error, the commit would fail. Regards, Rastislav ________________________________ From: Joan Landry Sent: Monday, August 1, 2016 2:50 PM To: sysrepo-devel at sysrepo.org Subject: [sysrepo-devel] verification options When a user changes configuration objects to the candidate data store via a series of edit-config Netconf calls, after doing so issues a commit. Is there a mechanism provided by sysrepo to allow the device to reject the commit if the combination of changes would result in an invalid configuration of the device? Or does sysrepo just automatically commit the changes per the request? I was not able to find any info on this - would greatly appreciate any info you could provide on how the device can verify the datastore before it is allowed to be commited? Thanks, Joan Landry RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr?la 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] -------------- next part -------------- An HTML attachment was scrubbed... URL: From JoLandry at advaoptical.com Mon Aug 1 13:01:46 2016 From: JoLandry at advaoptical.com (Joan Landry) Date: Mon, 1 Aug 2016 13:01:46 +0000 Subject: [sysrepo-devel] verification options In-Reply-To: <1470056055115.7797@pantheon.tech> References: <89475774d3ee4888827638c2cbd47f39@advaoptical.com> <1470056055115.7797@pantheon.tech> Message-ID: <224bcb3c71d341f080c1fd7885cc63d1@advaoptical.com> Thanks, Do you know the time frame for when this might be available as it would seem to be a mandatory piece in order to avoid a system with invalid configuration. Thanks for the quick response. Joan From: Rastislav Szab? [mailto:rastislav.szabo at pantheon.tech] Sent: Monday, August 01, 2016 8:54 AM To: Joan Landry ; sysrepo-devel at sysrepo.org Subject: Re: verification options Hi Joan, We do plan to support this, but it is not yet implemented. Currently sysrepo supports only post-commit notifications, meaning that callback is called AFTER the change has happened. Thi is signalled in the callback as event == SR_EV_NOTIFY. Any errors returned from the callback are ignored in this case. In the future, we would like to support callbacks being called even before the change has happened. This would require to specify SR_SUBSCR_VERIFIER flag in the subscription and it will be signalled as event == SR_EV_VERIFY in the callback. If that callback returns an error, the commit would fail. Regards, Rastislav ________________________________ From: Joan Landry > Sent: Monday, August 1, 2016 2:50 PM To: sysrepo-devel at sysrepo.org Subject: [sysrepo-devel] verification options When a user changes configuration objects to the candidate data store via a series of edit-config Netconf calls, after doing so issues a commit. Is there a mechanism provided by sysrepo to allow the device to reject the commit if the combination of changes would result in an invalid configuration of the device? Or does sysrepo just automatically commit the changes per the request? I was not able to find any info on this - would greatly appreciate any info you could provide on how the device can verify the datastore before it is allowed to be commited? Thanks, Joan Landry RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr??a 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] -------------- next part -------------- An HTML attachment was scrubbed... URL: From rastislav.szabo at pantheon.tech Mon Aug 1 13:13:14 2016 From: rastislav.szabo at pantheon.tech (=?iso-8859-1?Q?Rastislav_Szab=F3?=) Date: Mon, 1 Aug 2016 13:13:14 +0000 Subject: [sysrepo-devel] verification options In-Reply-To: <224bcb3c71d341f080c1fd7885cc63d1@advaoptical.com> References: <89475774d3ee4888827638c2cbd47f39@advaoptical.com> <1470056055115.7797@pantheon.tech>, <224bcb3c71d341f080c1fd7885cc63d1@advaoptical.com> Message-ID: <1470057194516.58925@pantheon.tech> Hi, a system with invalid configuration can be also avoided by a correct YANG model with constraints that would guard e.g. missing dependent data or wrong values (values out of range, or with wrong pattern). Sysrepo already supports this type of validation. But yes, real-time validation (not tied with the YANG model) is also an important part of the datastore. I cannot commit to any date, since we are now focusing on operational data support that is even more needed. But thanks for asking for that, we can prioritize it over other subsequent tasks. Reagrds, Rastislav ________________________________ From: Joan Landry Sent: Monday, August 1, 2016 3:01 PM To: Rastislav Szab?; sysrepo-devel at sysrepo.org Subject: RE: verification options Thanks, Do you know the time frame for when this might be available as it would seem to be a mandatory piece in order to avoid a system with invalid configuration. Thanks for the quick response. Joan From: Rastislav Szab? [mailto:rastislav.szabo at pantheon.tech] Sent: Monday, August 01, 2016 8:54 AM To: Joan Landry ; sysrepo-devel at sysrepo.org Subject: Re: verification options Hi Joan, We do plan to support this, but it is not yet implemented. Currently sysrepo supports only post-commit notifications, meaning that callback is called AFTER the change has happened. Thi is signalled in the callback as event == SR_EV_NOTIFY. Any errors returned from the callback are ignored in this case. In the future, we would like to support callbacks being called even before the change has happened. This would require to specify SR_SUBSCR_VERIFIER flag in the subscription and it will be signalled as event == SR_EV_VERIFY in the callback. If that callback returns an error, the commit would fail. Regards, Rastislav ________________________________ From: Joan Landry > Sent: Monday, August 1, 2016 2:50 PM To: sysrepo-devel at sysrepo.org Subject: [sysrepo-devel] verification options When a user changes configuration objects to the candidate data store via a series of edit-config Netconf calls, after doing so issues a commit. Is there a mechanism provided by sysrepo to allow the device to reject the commit if the combination of changes would result in an invalid configuration of the device? Or does sysrepo just automatically commit the changes per the request? I was not able to find any info on this - would greatly appreciate any info you could provide on how the device can verify the datastore before it is allowed to be commited? Thanks, Joan Landry RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr?la 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr?la 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] -------------- next part -------------- An HTML attachment was scrubbed... URL: From JoLandry at advaoptical.com Mon Aug 1 14:02:30 2016 From: JoLandry at advaoptical.com (Joan Landry) Date: Mon, 1 Aug 2016 14:02:30 +0000 Subject: [sysrepo-devel] verification options In-Reply-To: <1470057194516.58925@pantheon.tech> References: <89475774d3ee4888827638c2cbd47f39@advaoptical.com> <1470056055115.7797@pantheon.tech>, <224bcb3c71d341f080c1fd7885cc63d1@advaoptical.com> <1470057194516.58925@pantheon.tech> Message-ID: Thanks, possibly I will look to see if there is a way to intercede the copy-config command as I think if this one command had a pre verify callback ability that would go a long way to preventing a system level misconfiguration of the saved data store. Thanks again, Joan From: Rastislav Szab? [mailto:rastislav.szabo at pantheon.tech] Sent: Monday, August 01, 2016 9:13 AM To: Joan Landry ; sysrepo-devel at sysrepo.org Subject: Re: verification options Hi, a system with invalid configuration can be also avoided by a correct YANG model with constraints that would guard e.g. missing dependent data or wrong values (values out of range, or with wrong pattern). Sysrepo already supports this type of validation. But yes, real-time validation (not tied with the YANG model) is also an important part of the datastore. I cannot commit to any date, since we are now focusing on operational data support that is even more needed. But thanks for asking for that, we can prioritize it over other subsequent tasks. Reagrds, Rastislav ________________________________ From: Joan Landry > Sent: Monday, August 1, 2016 3:01 PM To: Rastislav Szab?; sysrepo-devel at sysrepo.org Subject: RE: verification options Thanks, Do you know the time frame for when this might be available as it would seem to be a mandatory piece in order to avoid a system with invalid configuration. Thanks for the quick response. Joan From: Rastislav Szab? [mailto:rastislav.szabo at pantheon.tech] Sent: Monday, August 01, 2016 8:54 AM To: Joan Landry >; sysrepo-devel at sysrepo.org Subject: Re: verification options Hi Joan, We do plan to support this, but it is not yet implemented. Currently sysrepo supports only post-commit notifications, meaning that callback is called AFTER the change has happened. Thi is signalled in the callback as event == SR_EV_NOTIFY. Any errors returned from the callback are ignored in this case. In the future, we would like to support callbacks being called even before the change has happened. This would require to specify SR_SUBSCR_VERIFIER flag in the subscription and it will be signalled as event == SR_EV_VERIFY in the callback. If that callback returns an error, the commit would fail. Regards, Rastislav ________________________________ From: Joan Landry > Sent: Monday, August 1, 2016 2:50 PM To: sysrepo-devel at sysrepo.org Subject: [sysrepo-devel] verification options When a user changes configuration objects to the candidate data store via a series of edit-config Netconf calls, after doing so issues a commit. Is there a mechanism provided by sysrepo to allow the device to reject the commit if the combination of changes would result in an invalid configuration of the device? Or does sysrepo just automatically commit the changes per the request? I was not able to find any info on this - would greatly appreciate any info you could provide on how the device can verify the datastore before it is allowed to be commited? Thanks, Joan Landry RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr??a 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr??a 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] -------------- next part -------------- An HTML attachment was scrubbed... URL: From JoLandry at advaoptical.com Tue Aug 2 13:54:23 2016 From: JoLandry at advaoptical.com (Joan Landry) Date: Tue, 2 Aug 2016 13:54:23 +0000 Subject: [sysrepo-devel] rpc lock and unlock Message-ID: <89c0014b68cb45d48b938be3f617bc41@advaoptical.com> Do you know if rpc-lock and unlock are supported as I get error when I try to lock any of the data stores. Locking datastore candidate by session 1 failed (Operation not authorized). Also, Is there a way to register for a callback on a lock and unlock to allow the device system level control over locking the config as there might be a non netconf interface that is modifying the config as well. Thanks, Joan -------------- next part -------------- An HTML attachment was scrubbed... URL: From rastislav.szabo at pantheon.tech Tue Aug 2 14:04:17 2016 From: rastislav.szabo at pantheon.tech (=?iso-8859-1?Q?Rastislav_Szab=F3?=) Date: Tue, 2 Aug 2016 14:04:17 +0000 Subject: [sysrepo-devel] rpc lock and unlock In-Reply-To: <89c0014b68cb45d48b938be3f617bc41@advaoptical.com> References: <89c0014b68cb45d48b938be3f617bc41@advaoptical.com> Message-ID: <1470146657065.58254@pantheon.tech> Hi, yes, lock and unlock should work. Check whether the user connecting to the NETCONF server has write permissions to the model you are trying to lock (sysrepoctl -l), as well as whether Netopeer2 and sysrepod are running under root privileges. As for the non-netconf interfaces that need to control the locks: cannot these just lock the datastore using sr_lock_datastore or sr_lock_module API when they are modifying the config? I don't see the use case for callback notifications upon lock and unlock, but if you think that it would be helpful, we can consider it. Could you please be a little bit more specific how they would be supposed to be used? Thanks, Rastislav ________________________________ From: Joan Landry Sent: Tuesday, August 2, 2016 3:54 PM To: sysrepo-devel at sysrepo.org Subject: [sysrepo-devel] rpc lock and unlock Do you know if rpc-lock and unlock are supported as I get error when I try to lock any of the data stores. Locking datastore candidate by session 1 failed (Operation not authorized). Also, Is there a way to register for a callback on a lock and unlock to allow the device system level control over locking the config as there might be a non netconf interface that is modifying the config as well. Thanks, Joan RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr?la 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] -------------- next part -------------- An HTML attachment was scrubbed... URL: From JoLandry at advaoptical.com Tue Aug 2 14:29:37 2016 From: JoLandry at advaoptical.com (Joan Landry) Date: Tue, 2 Aug 2016 14:29:37 +0000 Subject: [sysrepo-devel] rpc lock and unlock In-Reply-To: <1470146657065.58254@pantheon.tech> References: <89c0014b68cb45d48b938be3f617bc41@advaoptical.com> <1470146657065.58254@pantheon.tech> Message-ID: <50baa0ac3c1c441aa1cedd185a44baf1@advaoptical.com> I do not see a command to allow permission to lock a datastore. I am able to edit the yang model so the permissions on it are 666 - but I think this is not the same as permissions for locking the data store. The command I am trying to execute is From: Rastislav Szab? [mailto:rastislav.szabo at pantheon.tech] Sent: Tuesday, August 02, 2016 10:04 AM To: Joan Landry ; sysrepo-devel at sysrepo.org Subject: Re: rpc lock and unlock Hi, yes, lock and unlock should work. Check whether the user connecting to the NETCONF server has write permissions to the model you are trying to lock (sysrepoctl -l), as well as whether Netopeer2 and sysrepod are running under root privileges. As for the non-netconf interfaces that need to control the locks: cannot these just lock the datastore using sr_lock_datastore or sr_lock_module API when they are modifying the config? I don't see the use case for callback notifications upon lock and unlock, but if you think that it would be helpful, we can consider it. Could you please be a little bit more specific how they would be supposed to be used? Thanks, Rastislav ________________________________ From: Joan Landry > Sent: Tuesday, August 2, 2016 3:54 PM To: sysrepo-devel at sysrepo.org Subject: [sysrepo-devel] rpc lock and unlock Do you know if rpc-lock and unlock are supported as I get error when I try to lock any of the data stores. Locking datastore candidate by session 1 failed (Operation not authorized). Also, Is there a way to register for a callback on a lock and unlock to allow the device system level control over locking the config as there might be a non netconf interface that is modifying the config as well. Thanks, Joan RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr??a 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] -------------- next part -------------- An HTML attachment was scrubbed... URL: From rastislav.szabo at pantheon.tech Tue Aug 2 14:38:15 2016 From: rastislav.szabo at pantheon.tech (=?iso-8859-1?Q?Rastislav_Szab=F3?=) Date: Tue, 2 Aug 2016 14:38:15 +0000 Subject: [sysrepo-devel] rpc lock and unlock In-Reply-To: <50baa0ac3c1c441aa1cedd185a44baf1@advaoptical.com> References: <89c0014b68cb45d48b938be3f617bc41@advaoptical.com> <1470146657065.58254@pantheon.tech>, <50baa0ac3c1c441aa1cedd185a44baf1@advaoptical.com> Message-ID: <1470148695309.54333@pantheon.tech> The thing is, that the RPC you are sending via NETCONF is locking whole candidate datastore (= all modules), which means that you need to have write permissions into all modules installed in sysrepo. If you have write permissions on all the modules, and both Netopper2 and sysrepod are running under root privileges, than this probably will be a bug. In that case, please try to run both netopeer2-server and sysrepod in debug mode, (-d option), collect the logs and open the issue on GitHub. Thanks, Rastislav ________________________________ From: Joan Landry Sent: Tuesday, August 2, 2016 4:29 PM To: Rastislav Szab?; sysrepo-devel at sysrepo.org Subject: RE: rpc lock and unlock I do not see a command to allow permission to lock a datastore. I am able to edit the yang model so the permissions on it are 666 - but I think this is not the same as permissions for locking the data store. The command I am trying to execute is From: Rastislav Szab? [mailto:rastislav.szabo at pantheon.tech] Sent: Tuesday, August 02, 2016 10:04 AM To: Joan Landry ; sysrepo-devel at sysrepo.org Subject: Re: rpc lock and unlock Hi, yes, lock and unlock should work. Check whether the user connecting to the NETCONF server has write permissions to the model you are trying to lock (sysrepoctl -l), as well as whether Netopeer2 and sysrepod are running under root privileges. As for the non-netconf interfaces that need to control the locks: cannot these just lock the datastore using sr_lock_datastore or sr_lock_module API when they are modifying the config? I don't see the use case for callback notifications upon lock and unlock, but if you think that it would be helpful, we can consider it. Could you please be a little bit more specific how they would be supposed to be used? Thanks, Rastislav ________________________________ From: Joan Landry > Sent: Tuesday, August 2, 2016 3:54 PM To: sysrepo-devel at sysrepo.org Subject: [sysrepo-devel] rpc lock and unlock Do you know if rpc-lock and unlock are supported as I get error when I try to lock any of the data stores. Locking datastore candidate by session 1 failed (Operation not authorized). Also, Is there a way to register for a callback on a lock and unlock to allow the device system level control over locking the config as there might be a non netconf interface that is modifying the config as well. Thanks, Joan RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr?la 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr?la 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] -------------- next part -------------- An HTML attachment was scrubbed... URL: From mvasko at cesnet.cz Wed Aug 3 09:54:52 2016 From: mvasko at cesnet.cz (=?utf-8?q?Michal_Va=C5=A1ko?=) Date: Wed, 03 Aug 2016 11:54:52 +0200 Subject: [sysrepo-devel] pkg-config variable Message-ID: <7eb5-57a1bf80-b9-19c581c0@180653441> Hi, I was wondering whether you could add a CMake variable into libsysrepo.pc. For automatedsysrepo plugin installation the plugin directory needs to be detected somehow. It would be nice if PLUGINS_DIR could be retrieved using pkg-config, what do you think? Maybe some other variables could be relevant outside sysrepo installation, but I am interested in this one. Regards, Michal From rastislav.szabo at pantheon.tech Wed Aug 3 10:22:17 2016 From: rastislav.szabo at pantheon.tech (=?utf-8?B?UmFzdGlzbGF2IFN6YWLDsw==?=) Date: Wed, 3 Aug 2016 10:22:17 +0000 Subject: [sysrepo-devel] pkg-config variable In-Reply-To: <7eb5-57a1bf80-b9-19c581c0@180653441> References: <7eb5-57a1bf80-b9-19c581c0@180653441> Message-ID: <1470219737799.13794@pantheon.tech> Hi, yes, good idea! Will do that and let you know. Rastislav ________________________________________ From: Michal Va?ko Sent: Wednesday, August 3, 2016 11:54 AM To: sysrepo-devel at sysrepo.org Subject: [sysrepo-devel] pkg-config variable Hi, I was wondering whether you could add a CMake variable into libsysrepo.pc. For automatedsysrepo plugin installation the plugin directory needs to be detected somehow. It would be nice if PLUGINS_DIR could be retrieved using pkg-config, what do you think? Maybe some other variables could be relevant outside sysrepo installation, but I am interested in this one. Regards, Michal _______________________________________________ sysrepo-devel mailing list sysrepo-devel at sysrepo.org http://lists.sysrepo.org/listinfo/sysrepo-devel RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr??a 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] From rastislav.szabo at pantheon.tech Thu Aug 4 07:21:26 2016 From: rastislav.szabo at pantheon.tech (=?iso-8859-2?Q?Rastislav_Szab=F3?=) Date: Thu, 4 Aug 2016 07:21:26 +0000 Subject: [sysrepo-devel] pkg-config variable In-Reply-To: <1470219737799.13794@pantheon.tech> References: <7eb5-57a1bf80-b9-19c581c0@180653441>, <1470219737799.13794@pantheon.tech> Message-ID: <1470295285418.36491@pantheon.tech> Hi, I have exposed two variables via pkg-config: SR_REPOSITORY_LOC SR_PLUGINS_DIR The change has been already committed to the master branch. Thanks, Rastislav ________________________________________ From: Rastislav Szab? Sent: Wednesday, August 3, 2016 12:22 PM To: Michal Va?ko; sysrepo-devel at sysrepo.org Subject: Re: [sysrepo-devel] pkg-config variable Hi, yes, good idea! Will do that and let you know. Rastislav ________________________________________ From: Michal Va?ko Sent: Wednesday, August 3, 2016 11:54 AM To: sysrepo-devel at sysrepo.org Subject: [sysrepo-devel] pkg-config variable Hi, I was wondering whether you could add a CMake variable into libsysrepo.pc. For automatedsysrepo plugin installation the plugin directory needs to be detected somehow. It would be nice if PLUGINS_DIR could be retrieved using pkg-config, what do you think? Maybe some other variables could be relevant outside sysrepo installation, but I am interested in this one. Regards, Michal _______________________________________________ sysrepo-devel mailing list sysrepo-devel at sysrepo.org http://lists.sysrepo.org/listinfo/sysrepo-devel RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr??a 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] _______________________________________________ sysrepo-devel mailing list sysrepo-devel at sysrepo.org http://lists.sysrepo.org/listinfo/sysrepo-devel RastislavSzab? Senior Software Engineer S?dlo / Mlynsk? Nivy 56 / 821 05 Bratislava / Slovakia R&D centrum / Janka Kr??a 9 / 974 01 Bansk? Bystrica / Slovakia +421 918 462 530 / rastislav.szabo at pantheon.tech reception: +421 2 206 65 114 / www.pantheon.sk [logo] From lmacko at cisco.com Fri Aug 5 05:59:03 2016 From: lmacko at cisco.com (Lukas Macko -X (lmacko - PANTHEON TECHNOLOGIES at Cisco)) Date: Fri, 5 Aug 2016 05:59:03 +0000 Subject: [sysrepo-devel] subtree change default values References: <282-579a0680-1b-1fd77a60@37183671> Message-ID: <354b128e8af6488394cf9bca8b3c587e@XCH-ALN-015.cisco.com> Hi Michal, the change is merged in devel branch. If you set or delete existing default node the operation is SR_OP_MODIFIED and default flags are set. Regards, Lukas > -----Original Message----- > From: Lukas Macko -X (lmacko - PANTHEON TECHNOLOGIES at Cisco) > Sent: Thursday, July 28, 2016 3:59 PM > To: 'Michal Va?ko' ; sysrepo-devel at sysrepo.org > Subject: RE: [sysrepo-devel] subtree change default values > > Hi Michal, > > I agree that in the case you have described it would be more convinient to > recieve SR_OP_MODIFIED instead of created/deleted. > I'll look into it and I'll let you know. > > Regards, > Lukas > > > -----Original Message----- > > From: sysrepo-devel [mailto:sysrepo-devel-bounces at sysrepo.org] On > > Behalf Of Michal Va?ko > > Sent: Thursday, July 28, 2016 3:19 PM > > To: sysrepo-devel at sysrepo.org > > Subject: [sysrepo-devel] subtree change default values > > > > Hi, > > I started using sr_subtree_change_subscribe() and after very brief > > testing it works fine. However, when I iterate over the changes in the > > associated callback using sr_get_changes_iter() and > > sr_get_change_next(), I have never received any default values. Example: > > > > - leaf A with default value 50 > > - data include leaf A with value 20 > > - I remove the value > > > > What I get: > > - operation = SR_OP_DELETED > > - old_value = 20 > > - new_value = NULL > > > > What I wanted: > > - operation = SR_OP_DELETED > > - old_value = 20 > > - new_value = 50, dflt flag = true > > > > The same principle applied when creating the value as well, old_value > > = 50 with dflt flag. Also, it would maybe make more sense to change > > the operation to SR_OP_MODIFIED if there is a default value and the value is > created/removed. > > What do you think? > > > > Regards, > > Michal > > > > _______________________________________________ > > sysrepo-devel mailing list > > sysrepo-devel at sysrepo.org > > http://lists.sysrepo.org/listinfo/sysrepo-devel From rkrejci at cesnet.cz Wed Aug 10 13:28:36 2016 From: rkrejci at cesnet.cz (=?UTF-8?B?UmFkZWsgS3JlasSNw60=?=) Date: Wed, 10 Aug 2016 15:28:36 +0200 Subject: [sysrepo-devel] libyang and sysrepo performance Message-ID: <38a6508b-66c2-c5a9-592a-a27ae9b502fd@cesnet.cz> Hi Luka, since you mentioned today that you are planning some performance tests, check what is available at [1]. Maybe you could use the test cases and test schema from there. Later, maybe we could ask them to add sysrepo into their table :) The title "YumaPro SDK Transactions Over 350 Times Faster Than Open Source" irritates me, since they compare it only to OpenYuma and I believe that sysrepo with libyang can be faster at least in some of the test cases. [1] - https://www.yumaworks.com/netconfd-pro/#Transaction-Performance Regards, Radek From raszabo at cisco.com Wed Aug 10 14:03:07 2016 From: raszabo at cisco.com (Rastislav Szabo -X (raszabo - PANTHEON TECHNOLOGIES at Cisco)) Date: Wed, 10 Aug 2016 14:03:07 +0000 Subject: [sysrepo-devel] libyang and sysrepo performance In-Reply-To: <38a6508b-66c2-c5a9-592a-a27ae9b502fd@cesnet.cz> References: <38a6508b-66c2-c5a9-592a-a27ae9b502fd@cesnet.cz> Message-ID: <184a190e6e92403ea391b7ad065541ae@XCH-RCD-002.cisco.com> Good source Radek! Luka, we should definitely compare sysrepo with the same scenarios. My guess is: Test 1) We are not able to go over ~ 500 commits per second, since we are writing data to the filesystem with each commit. If they claim to be able to write 85,947 entries per second, they are definitely not doing that. We will be beaten here, but it would be interesting to see their results for target=startup. Test 2) I guess that after the recent tweaks that Radek did in libyang we will be definitelly much better than OpenYuma, close to YumaPro. Test 3) I have no idea since we have no perf unit-test for setting data yet :-( But I think that we will be more close to OpenYuma than YumaPro. Test 4) I believe that we could be very close to YumaPro. I can imagine that they tested use-cases they specifically had optimized for :-) but at least we have some baselines to compare with. Thanks, Rastislav -----Original Message----- From: sysrepo-devel [mailto:sysrepo-devel-bounces at sysrepo.org] On Behalf Of Radek Krej?? Sent: Wednesday, August 10, 2016 3:29 PM To: Luka Perkov Cc: sysrepo-devel at sysrepo.org Subject: [sysrepo-devel] libyang and sysrepo performance Hi Luka, since you mentioned today that you are planning some performance tests, check what is available at [1]. Maybe you could use the test cases and test schema from there. Later, maybe we could ask them to add sysrepo into their table :) The title "YumaPro SDK Transactions Over 350 Times Faster Than Open Source" irritates me, since they compare it only to OpenYuma and I believe that sysrepo with libyang can be faster at least in some of the test cases. [1] - https://www.yumaworks.com/netconfd-pro/#Transaction-Performance Regards, Radek _______________________________________________ sysrepo-devel mailing list sysrepo-devel at sysrepo.org http://lists.sysrepo.org/listinfo/sysrepo-devel From raszabo at cisco.com Fri Aug 19 14:16:58 2016 From: raszabo at cisco.com (Rastislav Szabo -X (raszabo - PANTHEON TECHNOLOGIES at Cisco)) Date: Fri, 19 Aug 2016 14:16:58 +0000 Subject: [sysrepo-devel] Contributing to sysrepo datastore Message-ID: <53bc419ed2294890b2d8b3d4e0cb9504@XCH-RCD-002.cisco.com> Hello everybody, many people from outside of the main development team of sysrepo expressed some interest in contributing to the sysrepo datastore codebase. For all potential volunteers, we created a WIKI page with some contribution rules [1] and put together a list of enhancements [2] and a TODO tasks [3]. For people who never touched sysrepo code, bugs and enhancements are best way to start. It is also a good idea to subscribe to sysrepo-devel at sysrepo.org mailing list [4] to be always up-to-date with the current development state and plans if you want to contribute. Thanks, Rastislav [1] https://github.com/sysrepo/sysrepo/wiki/Contributing [2] https://github.com/sysrepo/sysrepo/issues [3] https://github.com/sysrepo/sysrepo/wiki/TODO-List [4] http://lists.sysrepo.org/listinfo/ -------------- next part -------------- An HTML attachment was scrubbed... URL: From mislav.novakovic at sartura.hr Mon Aug 29 16:30:13 2016 From: mislav.novakovic at sartura.hr (Mislav Novakovic) Date: Mon, 29 Aug 2016 18:30:13 +0200 Subject: [sysrepo-devel] openssh sysrepo demo Message-ID: Hi everybody, we have integrated openssh with sysrepo using native sysrepo integration approach. That means that one can manage his openssh daemon via NETCONF and YANG ecosystem using sysrepo, Netopeer2 and libyang. In the example video we have used testconf suite from which we have sent NETCONF messages as NETCONF client would send. The documentation as well as video can be found at the sysrepo wiki: http://www.sysrepo.org/openssh-demo The source code with openssh sysrepo patches can be found on github: https://github.com/sysrepo/openssh-sysrepo When you try it out, if you have any questions about integration, please let me know. Best regards, Mislav Novakovic -------------- next part -------------- An HTML attachment was scrubbed... URL: