Just updated my Citadel to 996 and ....
citserver 38182 - - msgbase: message #8836 was not found
citserver 38182 - - msgbase: message #8837 was not found
citserver 38182 - - msgbase: message #8838 was not found
citserver 38182 - - msgbase: message #8839 was not found
citserver 38182 - - msgbase: message #8841 was not found
citserver 38182 - - msgbase: message #8842 was not found
citserver 38182 - - msgbase: message #8844 was not found
citserver 38182 - - msgbase: message #8845 was not found
citserver 38182 - - msgbase: message #8847 was not found
citserver 38182 - - msgbase: message #8848 was not found
Plus loads moar. I tried using the database_cleanup.sh and I still get this issue.
If you can figure out what room they are in, you can safely delete them.
To be clear: anyone who is running Citadel 993 or later should almost never have any need to run the database_cleanup script. I spent my whole summer making the database_bdb module bullet proof, and would be quite surprised if anyone ever had that kind of corruption again.
Now if you're *really* picky and have a lot of time to kill, you could run `ctdldump` and edit the dump by hand, and then `ctdlload` it back into a fresh database. But you've probably got better ways to spend your time.
The broken messages aren't in a room. They're in the Outbound SMTP Queue.
Mon Oct 09 2023 13:52:33 EDT from IGnatius T FoobarIt's possible that during some previous event those messages were deleted but one or more rooms still has pointers to them. In the new database code this should theoretically never happen, but if you still have those pointers to messages that don't exist, no amount of database cleanup will make the warnings go away.
If you can figure out what room they are in, you can safely delete them.
To be clear: anyone who is running Citadel 993 or later should almost never have any need to run the database_cleanup script. I spent my whole summer making the database_bdb module bullet proof, and would be quite surprised if anyone ever had that kind of corruption again.
Now if you're *really* picky and have a lot of time to kill, you could run `ctdldump` and edit the dump by hand, and then `ctdlload` it back into a fresh database. But you've probably got better ways to spend your time.
I'm attempting to upgrade our Citadel instance from version 958 to 996, using easyinstall (it was originally built using easyinstall).
The server is a 6-core, 12 thread I5-10400 with 64GB of RAM and NVMe disk, running Ubuntu 22.04.3 LTS 6.2.0-34-generic kernel.
The database size is 5.5GB, but has never had any issues. The server logs look clean; starting with 'citserver -x9' doesn't show any obvious problems.
But the update attempt always gets stuck (until I kill it after an hour) at:
Restarting Citadel server to apply changes
[************************************************************************....]
using 100% of one CPU core, but not generating disk I/O. So it's obviously getting stuck near the end of citadel/utils/setup.c
An strace on the setup process shows continual attempts of:
read(-1, 0x7ffcca70371f, 1) = -1 EBADF (Bad file descriptor)
(the read address changes from run to run)
Is this a repeated socket read attempt?
To proceed I have to kill the setup process and restore from backup.
Suggestions for how to diagnose further will be gratefully received...
The broken messages aren't in a room. They're in the Outbound SMTP
Queue.
Oh, that makes it easy then. The SMTP queue is in a hidden room called __CitadelSMTPspoolout__
Each item in the queue contains two messages: the message itself, and a set of delivery instructions. You can easily clean the queue using the text mode client.
Subject: Re: easyinstall upgrade from 958 to 996 hanging
To proceed I have to kill the setup process and restore from backup.
Have you tried just restarting the system without restoring from backup?
At that point, the upgrade has completed and you don't really need to complete the setup program if you aren't making any changes.
Also, knowing whether Citadel Server does or does not run correctly at that point might tell us what might have gone wrong. If it's just a setup problem that's one thing; if your server is non functional that's a different kind of problem.
Subject: Re: easyinstall upgrade from 958 to 996 hanging
I did try this once, on Sunday 1 October.
I stopped Citadel at 6:05pm and took a backup (tar -I pigz -cf /Backups/Citadel_bkp.tgz etc/fetchmailrc usr/local/citadel usr/local/webcit usr/local/ctdlsupport ).
I did a full server reboot at 6:23pm, then ran the update - at 6:30pm /var/log/syslog showed Citadel server engine starting with Version 996 (build 23274) and Existing database version on disk is 958, so it did the Running pre-startup and post-startup database upgrades. This took about one minute, leaving the database in a shut down state.
After letting the update script itself sit hanging for an hour, I Ctrl-C'd the update, then manually started citadel (systemctl start citadel.service) at 7:36pm
/var/log/syslog showed Citadel server engine starting with Version 996 (build 23274) and Existing database version on disk is 996
Things initially appeared to be OK - I received a test email at 7:48pm.
Then our normal scheduled backup ran at 8:10pm, stopping and restarting citserver. /var/log/syslog recorded normal database shutdown and restart.
But on Monday morning the filesystem was full due to 5,113 x 10MB log.00000nnnnn files in /usr/local/citadel/data/ (filled at 2:20am, but no citserver messages in /var/log/mail.log).
So I had to rollback to Saturday evening's backup taken immediately before I attempted the update.
Subject: Debian 12 Citadel Install
Hello,
I'm trying to setup citadel so that it works perfectly on a new box, Debian 12, but everytime i want to create a new room and setup the mailing list in order to have abuse@ or postmaster@ or admin@ or any other universal id such as hostmaster@ of a domain, the server simply refuses to save or adds ackward special characters to the domain name such as "%2%" and so on.
Do you know if there's a problem with debian 12 or am i doing something wrong?
I installed the server with the easy install script and there where some warnings but not errors.
Best Regards,
Niko
Subject: Re: Debian 12 Citadel Install
I think the issue might be the hyphen " - " from the domain name.
Can anybody please tell me if citadel has issues with domain names that contain hyphens?
BR,
Niko
Okay so basically I just need to start fresh with a new database. This thing is so borked there is no recovering it.
You can definitely use `imapsync` to copy important data from your old server instance to your new one. I've done that and it works quite well.
You can also use `ctdldump` and `ctdlload` for full fidelity -- but that *will* copy over the bad pointers so you probably don't want to do that.
Do you know if there's a problem with debian 12 or am i doing
something wrong?
There is definitely nothing wrong with Citadel running on Debian 12. If it runs at all, it built fine.
Please tell us exactly what you were trying to do and what steps you performed to get there.
Where can I find the log files (or is it possible to make citadel/webcit log to syslog)?
Could be if youre trying to authenticate with an active directory server and either you cant reach it, dont have the admin user created on the AD server, or something of the sort.
Subject: Active Directory Integration: Created emails and etc.
Hello All!
I have installed Citadel using easy install on ubuntu server 20.04.06.
During easy install I configured it to use Active Directory to authenticate users with. I noticed one thing in particular, when a user logs in using their login credentials, the user email that gets created uses the format "firstname_lastname@domain.com" instead of what I would expect to be "LoginName@domain.com". I believe it is using either the display name or the first and last name attributes in AD when creating a user. How can I change the mappings of AD attributes to change the created emails to use the sAMAccountName attribute?
I appreciate any insight!
Hi,
I was wondering if there was a file size restriction on incoming mail via citadel? reason I ask - a client sent me 11mb PDF document.. I failed to receive the attachment.. only the email appeared with no attachment.. after multiple tries I asked him to sent to gmail.. I received the file correctly.
please help.
Craig.
Subject: Re: File size restriction
From https://en.wikipedia.org/wiki/Email_attachment
Also note that all these size limits are based, not on the original file size, but the MIME-encoded copy. The common Base64 encoding adds about 37% to the original file size, meaning that an original 20MB file could exceed a 25MB file attachment limit.[13] A 10MB email size limit would require that the size of the attachment files is actually limited to about 7MB.
Hi,
I was wondering if there was a file size restriction on incoming mail via citadel? reason I ask - a client sent me 11mb PDF document.. I failed to receive the attachment.. only the email appeared with no attachment.. after multiple tries I asked him to sent to gmail.. I received the file correctly.
please help.
Craig.
Subject: citadel in docker container >> configuration issues
Hi, I'm trying to install a citadel/webcit server in a container (docker.-ce), but I'm finding it difficult to set up, for the smart-host on topic.
1) Oracle's mail delivery service (OCI) does not accept me.
2) I can't find where the Citadel configuration files are in the container (I would like to try setting the file manually).
Can anyone give me a hand?
Thanks!