> because it looks like the pmlogger service is actually being started on > separate issue not related to the systemd service issue reported in this BZ > to attach them to this BZ, could you mail them to me? This might be a > below /var/log/pcp/pmlogger and journalctl -b -u pmlogger. > to store more than one day of data? e.g. > Fernando, can you provide any more details about your system not being able > (In reply to Fernando Viñan-Cano from comment #1) (In reply to Mark Goodwin from comment #3) If you don't want to attach them to this BZ, could you mail them to me? This might be a separate issue not related to the systemd service issue reported in this BZ because it looks like the pmlogger service is actually being started on subsequent attempts by the PCP scripts, despite all the systemd complaints about the PID files etc. compressed tarball of your logs below /var/log/pcp/pmlogger and journalctl -b -u pmlogger. > What it does effect is the server is unable to store any more than a day ofįernando, can you provide any more details about your system not being able to store more than one day of data? e.g. > SELinux is not reporting any blocks, and then I had this disabled with F29 > new Fedora 30 machine the issue continues. > Fedora 29, then when upgraded to Fedora 30 it continued. > I can report the same issue, it was happening when my old box was running (In reply to Fernando Viñan-Cano from comment #1) Exactly what changes to make is yet to be determined. So I guess for PCP we're going to need to update the service unit files for all our services to be compatible with the newer versions of systemd. An example of an updated service file to comply with the new systemd is BZ #1583159. The end result is that /run/PID files should be owned by root and symlinks should not be used. As a result, systemd was subsequently updated to not follow symlinks, which apparently circumvented the original issue. xinetd and other services, see for example ). The change subsequently resulted in a spate of issues with various service daemons unable to start correctly because their PID file was not owned by root (e.g. The upstream systemd RFE requesting this is and the change that introduced the PID ownership check is. This issue seems to be similar to BZ #1583159 - "Sample service file is out of date with systemd changes", which was caused by an update to systemd in response to an upstream issue stating that PID run files should be owned by root, not an unprivileged user (such as 'pcp' in our case). I don't really know if there is anything wrong with the data, but I assume it is fine. I guess you need to have PCP services enabled and running, though. No action is required to produce this error. Version-Release number of selected component (if applicable):ġ. How can I fix the service so that it doesn't give errors every day? We'll most likely not notice when it exits. Jun 17 00:11:46 systemd: rvice: Supervising process 4066656 which Jun 17 00:11:38 systemd: Failed to start Performance Metrics Archive Logger. Jun 17 00:11:38 systemd: rvice: Failed with result 'timeout'. Jun 17 00:11:38 systemd: rvice: Start operation timed out. Jun 17 00:10:13 systemd: rvice: New main PID 4059499 does notīelong to service, and PID file is not owned by root.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |