*** Not <Not!~notifico@www.globalqss.com> has joined #idempiere | 02:45 | |
Not | [iDempiere] 1 commit to default [+5/-0/±1] https://bitbucket.org/idempiere/idempiere/commits/ | 02:45 |
---|---|---|
Not | [iDempiere] hengsin 9083c2d - IDEMPIERE-4060 Implement change scheduler state at scheduler window. Deactivate toolbar button and add button field implementation | 02:45 |
Not | [iDempiere] jenkins build #198 started http://ci.idempiere.org/job/iDempiere/198/ | 02:51 |
Not | [IDEMPIERE] Heng Sin Low updated IDEMPIERE-4051 assignee set to "Elaine Tan" | 02:57 |
Not | [IDEMPIERE] http://idempiere.atlassian.net/browse/IDEMPIERE-4051 | 02:57 |
Not | [iDempiere] jenkins build #198 completed: success http://ci.idempiere.org/job/iDempiere/198/ | 02:59 |
Not | [IDEMPIERE] Elaine Tan updated IDEMPIERE-4051 Attachment set to "IDEMPIERE-4051_2.patch" | 04:18 |
Not | [IDEMPIERE] http://idempiere.atlassian.net/browse/IDEMPIERE-4051 | 04:18 |
*** nmicoud <nmicoud!~Nicolas@2a01:e34:eeb6:cb10:c866:ec51:cfba:7e32> has joined #idempiere | 04:30 | |
Not | [IDEMPIERE] Heng Sin Low updated IDEMPIERE-4051 assignee set to "Heng Sin Low" | 05:44 |
Not | [IDEMPIERE] http://idempiere.atlassian.net/browse/IDEMPIERE-4051 | 05:44 |
*** a42niem <a42niem!~niem@p578b7daf.dip0.t-ipconnect.de> has joined #idempiere | 06:13 | |
*** CarlosRuiz <CarlosRuiz!~CarlosRui@ip-95-223-106-12.hsi16.unitymediagroup.de> has joined #idempiere | 07:20 | |
*** a42niem <a42niem!~niem@p578b7daf.dip0.t-ipconnect.de> has quit IRC | 07:26 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has joined #idempiere | 07:53 | |
norbertbede | hi. | 07:54 |
norbertbede | can we call maintenance mode from api ? https://wiki.idempiere.org/en/NF2.0_Maintenance_Mode | 07:54 |
CarlosRuiz | Hi norbertbede - I guess it would be something like 1) update the sysconfig 2) cache reset 3) close all sessions | 07:57 |
norbertbede | we have done Kill session as process =>callable from SOAP client, we can define timeout but thinking add change maintenance mode to ='Y'. but still no process to set back maintenance mode to Y | 08:01 |
norbertbede | *sorry for 'N' | 08:02 |
CarlosRuiz | would be just the same I think - update the sysconfig - cache reset | 08:04 |
norbertbede | also we sending this per node - each server independently | 08:04 |
norbertbede | so we create another hardcoded endpoint to change a specific sysconfig Y/No. yes that looks reasonalbe | 08:05 |
norbertbede | let me ask last question: when we send pending restart - disallow login new users - we still cant wait for processes running - and users usually start until latest seconds processes. | 08:06 |
norbertbede | so the question is: is it legitimate add to processes: option: if maintenance mode = Y then cant start new process ? | 08:07 |
norbertbede | or disallow create new records - is that something weird ? | 08:07 |
norbertbede | ATM we found some data consistency errors, e.g m_transactions in client=0 - so looks restart cant be done in days- but we have customer receiving orders each minut 6 days in week. so i need make clean picture, and solution for that | 08:08 |
CarlosRuiz | when a session is killed - all the processes running from that session are killed too, no? | 08:09 |
norbertbede | need deeper test. what if process commit 50% of actions into database, then cant rollback or we need refactor all processes | 08:09 |
norbertbede | would be better wait seriously and do not allow start new ones - probably we can rollout this instance from proxy. | 08:10 |
CarlosRuiz | a process must not commit 50% of a complete job - that would be very bad | 08:10 |
CarlosRuiz | I mean - there are processes that have the long transaction split in many mini-transactions - but every transaction is a complete work | 08:11 |
norbertbede | ok. here i need talk to coder. maybe i was wrong. | 08:11 |
norbertbede | i mean shipment generated for 300 lines, but lines are commited on each orderline. | 08:15 |
norbertbede | so physically created so cant rolled back anymore | 08:15 |
CarlosRuiz | if lines are committed on each orderline - that's an error - must commit just atomic transactions - all or nothing | 08:17 |
CarlosRuiz | if you have a process that commits partial stuff - then you need another process that clean partial stuff when you have an outage | 08:18 |
norbertbede | i see. so normally we need commit at end | 08:20 |
norbertbede | we check it. | 08:20 |
CarlosRuiz | or commit complete work - can be partial - but things that are complete | 08:21 |
CarlosRuiz | for example Generate Shipments can be refactored to commit when a complete shipment is generated - but not on lines | 08:21 |
norbertbede | so you mean the solution is to have proper - commit at end. then when shutdouwn done, simple kill and rollback processes | 08:23 |
norbertbede | another use case: docaction, complete document - we have an issue, when comete on doc with 100 lines duration 100 sec, user logout then lost context and m_transactions saved into ad_client_id=0 | 08:23 |
CarlosRuiz | I think we have a control about that in PO.java | 08:24 |
CarlosRuiz | PO.checkValidContext() | 08:25 |
norbertbede | probably we should talk at conference about that | 08:25 |
CarlosRuiz | IDEMPIERE-2756 | 08:25 |
CarlosRuiz | exactly the case you're describing - it was solved 4 years ago | 08:25 |
CarlosRuiz | https://idempiere.atlassian.net/browse/IDEMPIERE-2756 | 08:25 |
CarlosRuiz | you added a comment in the ticket | 08:26 |
norbertbede | ok. i will start these use cases. let you know - maybe another case | 08:26 |
CarlosRuiz | maybe you didn't integrate it? | 08:26 |
norbertbede | ok we check it. thanks for comments | 08:26 |
norbertbede | FYI: im start to play with ansible to allow automate and make correct approach to nodes rollin, rollout. toast instanced, propagate into proxy. | 08:36 |
norbertbede | would be intereseting how trek solved this. probably this is secure information | 08:36 |
norbertbede | cant imagine they do have or not automate their infrastructure | 08:36 |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has quit IRC | 08:53 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has joined #idempiere | 08:55 | |
norbertbede | @carlosruiz. is it possible to disallow users to pin topics in forum - Display at the top ? would be better to dissallow set this option to regular users, and unpin existing ones. | 09:42 |
CarlosRuiz | recently I unpinned a lot of messages - and now I see again 17 pinned messages | 10:09 |
CarlosRuiz | let me check if there is an option to forbid it | 10:09 |
norbertbede | ok. thanks would be really correct - sure important like conference, release is legitimate :) | 10:28 |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has quit IRC | 11:37 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has joined #idempiere | 11:38 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has quit IRC | 11:56 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has joined #idempiere | 11:56 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has quit IRC | 12:06 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has joined #idempiere | 12:07 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has quit IRC | 12:19 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has joined #idempiere | 14:05 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has quit IRC | 15:02 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has joined #idempiere | 15:06 | |
*** norbertbede <norbertbede!~Adium@94.228.89.101> has quit IRC | 15:20 | |
*** a42niem <a42niem!~niem@p578b7daf.dip0.t-ipconnect.de> has joined #idempiere | 16:58 | |
*** CarlosRuiz <CarlosRuiz!~CarlosRui@ip-95-223-106-12.hsi16.unitymediagroup.de> has quit IRC | 17:34 | |
*** nmicoud <nmicoud!~Nicolas@2a01:e34:eeb6:cb10:c866:ec51:cfba:7e32> has quit IRC | 18:55 | |
*** a42niem <a42niem!~niem@p578b7daf.dip0.t-ipconnect.de> has quit IRC | 20:48 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!