Hide Forgot
Created attachment 1191495 [details] messages Description of problem: there is an error log with rhel7.3 Version-Release number of selected component (if applicable): How reproducible: stable reproducible Steps to Reproduce: 1. Full install RHEL7.3Alpha1 on X3650M5 with UEFI mode 2. boot into OS and check log Actual results: error message Expected results: No error message; Additional info:
Hello Lenovo, Please attach a sosreport directly after seeing this issue. Thank You Joe Kachuck
Hi Joe Kachuck： The server is in use by others; I will catch the log tomorrow; Thanks
Hi: I couldn't reproduce this issue,I will continue try to work it out, if i failed to do so, I will close the bug for one off Thanks for your support;
During some sort of restart? Aug 11 00:09:01 localhost systemd: Stopping LVM2 PV scan on device 8:3... ... Aug 11 00:09:01 localhost systemd: Stopped LVM2 PV scan on device 8:3. Aug 11 00:09:01 localhost systemd: Stopping LVM2 metadata daemon... Aug 11 00:09:01 localhost systemd: Removed slice system-lvm2\x2dpvscan.slice. Aug 11 00:09:01 localhost systemd: Stopping system-lvm2\x2dpvscan.slice. Aug 11 00:09:01 localhost lvmetad[745]: Failed to accept connection. Aug 11 00:09:01 localhost systemd: Stopped LVM2 metadata daemon.
(In reply to Alasdair Kergon from comment #5) > During some sort of restart? > > Aug 11 00:09:01 localhost systemd: Stopping LVM2 PV scan on device 8:3... > ... > Aug 11 00:09:01 localhost systemd: Stopped LVM2 PV scan on device 8:3. > Aug 11 00:09:01 localhost systemd: Stopping LVM2 metadata daemon... > Aug 11 00:09:01 localhost systemd: Removed slice system-lvm2\x2dpvscan.slice. > Aug 11 00:09:01 localhost systemd: Stopping system-lvm2\x2dpvscan.slice. > Aug 11 00:09:01 localhost lvmetad[745]: Failed to accept connection. > Aug 11 00:09:01 localhost systemd: Stopped LVM2 metadata daemon. The sequence looks good - first all the clients are stopped (here there lvm2-pvscan@8:3.service), then the lvmetad daemon. Although, I don't see stopping ov lvm2-monitor.service in the log (which can also communicate with lvmetad), but that one has "After=lvm2-lvmetad.service" for start, so the other way round for the stop operation - so stop for lvm2-monitor goes before lvmetad stop operation so this should not pose a problem. There are no other our own services that may cause lvmetad to be contacted, so I suppose this is probably some other client - someone calling LVM tool directly during shutdown process... Now, the "Stopping LVM2 metadata daemon" means that systemd is sending SIGTERM for the process which in lvmetad causes the "_shutdown_requested" variable to be set as part of signal handler. Then the variable is checked after any last connection is handled and then the "_shutdown_requested" breaks the main lvmetad loop. So, any last request should be handled correctly without premature socket closing on which "accept" call would fail. We'd be wiser if we had the actual errno for the "accept" call - which we found missing in our code and we added that now for better debugging. If this is reproducible, I can provide an updated package with this extra debug added...
So in summary: 1) We have added logging of the error number that the system call returns so we will get more information if this happens again. 2) If you do find you can reproduce this, then we can supply you with an updated package now directly to provide the addtional diagnostics. Otherwise it'll be a few weeks before the change reaches a package you receive through the normal channels.
Hi Yonggang: have you ever reproduce this error log,"lvmetad[745]: Failed to accept connection"; Thanks a lot;
Hi： We haven't reproduced this issue； I think we can close it. if Lenovo find a way to reproduce this, we can reopen it; Is this OK? Thanks