support for setting detail level specific to fail2ban
I should add this fixes all dovecot parsing problems against the last week of logs on my smtp server.
fix some dovecot parsing and various other issues
Minor change to systemd output for skipped conditions
systemd 257.13: New skipped format
So I have the same problem with Chronyd: I have two laptops, on one I get a report from the chronyd service, probably because there is a Warning message in messages, on the second there is no entry for chronyd. I checked, and chronyd is definitely running. I ran logwatch --debug 100 and looked through all the output. There are entries in messages for the relevant day, but they are not warnings. Now, I looked through the chronyd service file and added: if ( $Debug >= 5 ) { print STDERR "\n\nDEBUG:...
Yes, it did work correctly the following day. I will try setting --debug 100 if it happens again
That indeed fixes it. Thanks!
As mentioned in an email to you on May 26, 2024, we are unable to merge and support this code into Logwatch, and we encourage you to spin it off as a separate Sourceforge project for those Logwatch users that would like to use XML/JSON formats. I am closing the merge request.
XML and JSON extension
Was merged, with some changes to comments. Thanks for the code, and for taking care that it does not break existing installations.
Add kernel-journal: journald-sourced kernel/dmesg service
New exim script pushed to repository; let us know if it fixes it. As per the link you provided, openssl reports it depending on its version. For this script, I've pushed it to only print if Detail is above 50, as is already the case for other log statements in this script.
[exim] Filtering non-properly terminated TLS connection; reported by Stamgastje.
[kernel-journal] Added kernel-journal, by Sourcekiller
[sssd] ignore single 0 log messages due to https://github.com/SSSD/sssd/issues/8928
Add kernel-journal: journald-sourced kernel/dmesg service
Add kernel-journal: journald-sourced kernel/dmesg service
Add kernel-journal: journald-sourced kernel/dmesg service
Issue with ignore.conf lines for exim
Did it work the next day? Or where there any changes in the environment that might have caused that? You can also set --debug 100. As mentioned in the man page, the temporary files created in /var/cache/logwatch will not be removed after the logwatch command completes. Examine those files to see if there are any statements that should have been captured by Logwatch. That debug level will also provide much detail on what is being executed. Remove those temporary files when no longer needed; otherwise...
No entry in today's log for iptables
[systemd] Coalescing start/reload/deactivate statements with different instance IDs.
[sshd] Suppressing output error due to deprecated logging, and
[sendmail] Relabeling incorrect 'error' to 'warning' for STARTTLS summary.
[dovecot] Allow additional 'Info' tag on 'Logged in' statement.
logwatch cron uses erroneous path in newer EL versions.
Works perfectly, thank you.
Thanks, Frank. Sorry for the late reply. I will install it and run it against my logs.
Sven, the updates I made are now available in the new release 7.15.
Andrew, this update is now available in the new release 7.15.
Terry, these updates are now available in the new release 7.15.
Preparation for release 7.15
[logwatch.cron,logwatch.service] Changing executable path to bin.
dovecot: Update to version 2.4
I'm rejecting this now, as it has been rolled into the recent update I just pushed.
Terry, as noted in the other discussion, I've kind of rolled this into the update now in the repo. Please check and confirm.
Andrew, something similar to your request has now been pushed to the repo. Sorry about the delay, but please check to see if it fixes your issue.
Folks, I've picked up items from both of these requests and pushed an update to dovecot for them. Please check if they address all the issues.
Merge outstanding fixes for dnf5 and dovecot.
Updates for dnf5 and dovecot
Updates for dnf5 and dovecot
Sven, I've cherry-picked your merge request to add it in with another request that we have had. I'll push it to the repo today and if you can check it works as you expect please let us know ASAP.
[fail2ban] Additional correction to force setting of service variable.
[rsyslogd] Ignore "epoll subsystem signaled EPOLLERR for stream" messages
[clam-upddate] Ignore_Outdated now ignores "DNS record is older" messages
[rsyslogd] handle updated not permmitted to talk to peer message; Only add unhandled connection closed messages to generic hash
[named] Ignore more key messages
[lvm] Handle VDO pools
[evtapplication] simplify Application Hang output
[rsyslogd] Handle "cannot resolve hostname"
[rsyslogd] Add rsyslogd_ignore_connection_closed_error_messages option to ignore more connnection closed error messages
[systemd] Ignore reload requested from systemctl
[system] Ignore Sending signal SIGUSR2 to process message
[systemd] Ignore sys.module.fuse.device message
[fail2ban] Filtering out prefix (date, service, PID) because prior git
Pushed change to /usr/bin rather than /usr/sbin as executable for logwatch (which is a symbolic link to the /usr/share/logwatch/scripts/logwatch.pl script). Does not delete any links under /usr/sbin/logwatch. For existing systems, the old link continues to work. For new installations, only the new /usr/bin/logwatch link will work. Also deleted the requirement for crontabs package, as newer installations may use alternate scheduling tools (most notably systemd timers).
[logwatch] Using /usr/bin rather than /usr/sbin as target directory.
[freeradius] Ignore "rlm_ldap: libldap vendor:" version message
dovecot 2.4: handle changed log format for aborted logins
Now I see that both change requests might have overlapped. So maybe submit it as a patch file from the prior dovecot or, even better, as either a patch file or a merge request from the current repository's version.
Can you re-submit the merge request with the new base? Your previous patch was already saved to the repository. So your working dovecot file should be the correct version, with the additional changes.
[sshd] Ignore connection resets during writes, per Martijn Pieters
SSHD network errors: connection reset by peer during write
Thanks for the patch.
There are also other changes, please see: https://sourceforge.net/p/logwatch/git/merge-requests/90/
SSHD network errors: connection reset by peer during write
There appears to be a push among some distributions/packages to move to /usr/bin, and eventually remove /usr/sbin. So my inclination would be to change all references and installation of logwatch to /usr/bin. In both the logwatch.spec and install_logwatch.sh file, we can test if the /usr/sbin directory exists. If so, then add a symbolic link from /usr/sbin/logwatch to /usr/bin/logwatch, for those that have hard-coded the path in their own scripts. That's not the end of the story, however. Some of...
dovecot 2.4: handle changed log format for aborted logins
Thanks for the changes.
System is tainted: unmerged-bin
Thanks for the report. Error is hidden unless $Detail is 10 or higher.
[systemd] Hide 'unmerged-bin' error if Detail level less than 10, based
Fix for sshd disconnect with no reason given
Thanks for the patch.
[sshd] Allow for disconnects listed without reason, per Reio Remma
[dovecot] Update due to dovecot 2.4 that uses "Login aborted",
[postfix] Corrected typo; submitted by Peter K.
I believe this is related/similar so replying here vs. creating a new bug. First install of v7.14 on a (new) RockyLinux 9.7 VM yesterday and received this error overnight: /etc/cron.daily/0logwatch: line 14: /usr/sbin/logwatch: No such file or directory Confirmed the file was missing: ls -lrt /usr/sbin/logwatch ls: cannot access '/usr/sbin/logwatch': No such file or directory On other RL8/RL9 installs using prior versions (both v7.9 and v7.13), this appears to be a link: ls -lrt /usr/sbin/logwatch...
[pam_unix] Simplify regest for gdm session opened, support users with . in names
named: Silence some informative messages and unhide a warning
Looks good to me. Thanks.
dovecot: Update to version 2.4
named: Silence some informative messages and unhide a warning
I think the following changes should suffice: diff --git a/logwatch/scripts/services/dovecot b/logwatch/scripts/services/dovecot index 5683be70..af8527ad 100755 --- a/logwatch/scripts/services/dovecot +++ b/logwatch/scripts/services/dovecot @@ -194,7 +194,7 @@ while (defined(my $ThisLine = <STDIN>)) { ($ThisLine =~ /ssl-params: Generating SSL parameters/) or ($ThisLine =~ /auth-worker/) or ($ThisLine =~ /auth:.*: Connected to/) or - ($ThisLine =~ /Disconnected: Connection closed(?! \(auth failed)/)...
dovecot 2.4: handle changed log format for aborted logins
[named] Update forwarded result NOERROR is not an error
System is tainted: unmerged-bin
If RHEL 9.x+ and children have changed the path of the logwatch script in the rpm spec (which is normally a symlink in any case) from sbin to bin, then whatever solution also needs handle all instances of the path…. A simple grep shows we have 4 files which have the hardcoded /usr/sbin/logwatch README:ln -s /usr/share/logwatch/scripts/logwatch.pl /usr/sbin/logwatch install_logwatch.sh:ln -f -s $BASEDIR/scripts/logwatch.pl /usr/sbin/logwatch install_logwatch.sh:printf "Created symlink for /usr/sbin/logwatch...
So maybe sed the path in 0logwatch replacing the path with sbindir variable during build. Just manually ran sed to get things working again: sed -i 's/LOGWATCH_SCRIPT="\/usr\/sbin\/logwatch"/LOGWATCH_SCRIPT="\/usr\/bin\/logwatch"/' /etc/cron.daily/0logwatch
logwatch cron uses erroneous path in newer EL versions.
Fix for sshd disconnect with no reason given
Sorry about that—I noticed this when I was looking at the Logwatch messages for version 7.14 this morning. Nothing like that is shown in the Dovecot section anymore; only the “Unmatched Entries” are displayed: --------------------- Dovecot Begin -------------------- ---- Unmatched Entries dovecot: imap-login: ID sent: : user=<>, rip=87.236.176.139, lip=1.2.3.4, TLS, session=<qcrscelp3exx7lcl>: 1 Time(s)</qcrscelp3exx7lcl> ---------------------- Dovecot End ------------------------- This feels a bit...
Similar issue to the one you posted for postfix: the output shown corresponds to an older version; there have been substantial changes to the dovecot script since your version. Let us know if the issue persists after upgrading to the current Logwatch version 7.14.
OK, thanks. I installed version 7.14 on my mail server and sent an email to a distribution list, but I can't find any record of it in the system logs. I have now also installed version 7.14 on the customer’s server where I discovered this “unmatched” message, and I’m checking to see if it now appears correctly in the watch log or no longer as “unmatched.” However, emails aren’t sent that often to distribution lists.
I’m happy to provide both here in anonymized form. I don’t know whether the data from the log file matches exactly what’s in the Logwatch output, but
Typo in message output