Filter Rules disappearing
-
You could try to setup a new installation/mailbox and see if you get the same results there as well.
Its not like we discard some data during upgrade, so I don’t think it would make sense to look in the server.log, maybe raise logging of the dagent and see if it logs something interesting when a rule should be applied.
-
I tried upgrading the core again with nightly from last night. The issue persists. I raised the log level to 6 in both dagent and server, configuration, but there is really nothing relevant in the log files. I can create new rules which are then shown in the Web App. However, all old rules (rules that were created before upgrading) are gone…
When creating a new mailbox - as suggested by you - there won’t be any rules as they are local to the user, or isn’t it like that?
-
@ashceryth said in Filter Rules disappearing:
When creating a new mailbox - as suggested by you - there won’t be any rules as they are local to the user, or isn’t it like that?
Yes, but I meant like “check if it’s something related to your inbox, or does it happen to other mailboxes as well”. For this you of course need to create a rule in the inbox before the upgrade.
-
Well, that makes sense now :) I will try that later and report the results. Very strange issue.
-
Maybe also interesting: The new rule that has been created after the upgrade for testing purposes has overwritten all old rules. So after restoring the KC VM snapshot, there was only this one rule so I had to restore the database as well.
-
@fbartels said in Filter Rules disappearing:
[…], anyone else experiencing it?
I observed a similar thing:
Webapp-settings got corrupted, entirely with one of the recent master-builds:- rules disappearing
- inbox-privileges were broken (no access)
- webapp started with initial setup-dialog
Re-creating the user fixed the problem for a short time, but corruption appeared again.
I didn’t report the issue, as I was too busy and not sure, if a build-problem was the main-cause on my system.
It happened with one of the recent master-builds (1- 2 weeks ago), but didn’t show up again, since.++umgfoin.
-
I tried to reproduce it (updated from 8.6.2 to 8.6.80 in core), but my rules weren’t affected by the upgrade…
-
After upgrading to the latest nightly yesterday all my rules got lost.
-
Hello @all
there is a script (https://stash.kopano.io/projects/KSC/repos/core-tools/browse/kopano-rules) which you can use to list all rules of a specific user. - Check if the rule is still shown.
coffee_is_life
-
@Coffee_is_life All gone.
-
@bmaehr and i bet dagent log doesnt show lines like
[info ] [32079] Rule <Some Rule_Name> doesn't match: 0x8004010fthere must be somethink broken with the listing and processing of these rules, since they exist in DB.
even the python-script cant see or locate these rulessince you are using the nightly builds, i dont think you got subsription (if i’m wrong, best way would be to open a ticket)
this issue is huge and got a greate impact on daily workflow, so i think the devs will look into this asap - with and without ticketcoffee_is_life
-
dagent log is completely empty (created 3 months ago).
That’s the bad thing in kopano using the SQL db just as data heap: I can’t check if they are really there. I don’t think they will come back, but perhaps I get surprised.
The rule created with core-8.6.80.1018 didn’t get deleted after upgrading to core-8.6.80.1055, so it seams not to be a problem of the upgrade itself but more a problem of the format of the old rules. -
Same problem here, at least with one User. Another Users rules dont seem to be affected.
Problematical user recreated his rules, are visible with py-script but no more in webapp and dont get applyed :(
-
Up until this point we were sadly not able to reproduce this issue. If you have any more information about the rules of the user, or with which master version this appeared first we would be happy to hear about this.
-
I’ve upgraded from core-8.6.80.752_0+85-Debian_9.0-amd64 to core-8.6.80.1018_0+152-Debian_9.0-amd64, then the problem occured.
The User was greeted by the webapp configuration wizard, others not.
Afterwards, i upgraded to core-8.6.80.1055_0+156-Debian_9.0-amd64 with no change.Then i recreated the rules in webapp, after switching windows in webapp they disappeared there again.
Using rules based on “Includes these words in the sender’s address” are showing up correctly in py-script, those using full senders address are having an empty Condition-Table entry. Maybe there the Problem comes from?
After Upgrade working users rules only use the first, working, condition.
Another thing i notices:
The problematic user is also missing the mail signature, setup via webapp. -
@fbartels Can you provide a SQL Statement to list all rules from the database directly? So we could check if they are still there.
-
After deleting the Rules with missing conditions, the remaining rules reappeeared in the Webapp, test if they’re working is pending.
-
Update: Not working and rules disappeared again :(
-
If you have any more information
Better yet, if you have any disk images, shell prompts, etc., …
SQL Statement to list all [dagent] rules from the database directly?
SELECT h.owner, val_binary FROM properties AS p INNER JOIN hierarchy AS h ON p.hierarchyid=h.id WHERE tag=0x3FE1 -
@jengelh Ok, I have executed the statement on my updated system and imported a database-dump from some months ago into an other database and executed the statement. I don’t think that I changed the rules since then - just created one rule after the update when all rules got lost.
Both are returning exactly one record with a binary field. The size of the field is very different: about 58kb on the old and about 2kb on the new.I’m not really able to read the content of the binary, but I was able to find the sender addresses I use with the rules to sort the mail into different folders inside the binary blob.
In the old one there are many addresses (I expect all addresses of all rules) in the blob and in the new one only one for the one rule I created.So it seams to me all rules get lost also in the database and/or the whole blob was thrown away.
Could I help you in some way to find the problem? Perhaps exporting my old blob in some way and sending it to you?
