Description of Problem: Aashu Virmani of Winphoria found a bug in some cases where literally thousands of /etc/tmp.XXXXXX files become created. Version-Release number of selected component (if applicable): 1.0.11 How Reproducible: I have not successfully reproduced the problem. Additional Information: This behavior is exhibited after the following behavior is noted in the logs: Jul 9 15:25:09 nmhost01 clusvcmgrd[5043]: <info> Starting stopped service wms_cb2 Jul 9 15:25:10 nmhost01 clusvcmgrd: [5044]: <notice> service notice: Starting service wms_cb2 ... Jul 9 15:25:10 nmhost01 clusvcmgrd: [5044]: <err> service error: 'mount -t ext3 -o rw,nosuid,sync /dev/sde6 /cluster/cdr/cdrcp02' failed, error=16 Jul 9 15:25:10 nmhost01 clusvcmgrd: [5044]: <err> service error: can't link lock file /etc/mtab~: No such file or directory (use -n flag to override) Jul 9 15:25:10 nmhost01 clusvcmgrd: [5044]: <err> service error: Cannot start filesystems for wms_cb2 Jul 9 15:25:10 nmhost01 clusvcmgrd: [5124]: <notice> service notice: Stopping service wms_cb2 ... Jul 9 15:25:10 nmhost01 clusvcmgrd: [5124]: <notice> service notice: Running user script '/opt/WMS/cluster/wms_cb2.sh stop' Jul 9 15:25:10 nmhost01 clulog[5157]: <info> Running user stop script for service wms_cb2 Jul 9 15:25:11 nmhost01 clusvcmgrd: [5124]: <info> service info: stty: standard input: Inappropriate ioctl for device Jul 9 15:25:11 nmhost01 clusvcmgrd: [5124]: <info> service info: stty: standard input: Inappropriate ioctl for device Jul 9 15:25:12 nmhost01 clusvcmgrd: [5124]: <info> service info: /dev/sde6 is not mounted Jul 9 15:25:14 nmhost01 clusvcmgrd: [5124]: <notice> service notice: Stopped service wms_cb2 ... Jul 9 15:28:54 nmhost01 clusvcmgrd[5590]: <info> Starting disabled service wms_cb2 Jul 9 15:28:54 nmhost01 clusvcmgrd: [5591]: <notice> service notice: Starting service wms_cb2 ... Jul 9 15:28:54 nmhost01 clusvcmgrd: [5591]: <err> service error: 'mount -t ext3 -o rw,nosuid,sync /dev/sde6 /cluster/cdr/cdrcp02' failed, error=32 Jul 9 15:28:54 nmhost01 clusvcmgrd: [5591]: <err> service error: mount: /dev/sde6 already mounted or /cluster/cdr/cdrcp02 busy Jul 9 15:28:55 nmhost01 clusvcmgrd: [5591]: <err> service error: Cannot start filesystems for wms_cb2 Jul 9 15:28:55 nmhost01 clusvcmgrd: [5670]: <notice> service notice: Stopping service wms_cb2 ... Jul 9 15:28:55 nmhost01 clusvcmgrd: [5670]: <notice> service notice: Running user script '/opt/WMS/cluster/wms_cb2.sh stop' Jul 9 15:28:55 nmhost01 clulog[5702]: <info> Running user stop script for service wms_cb2 Jul 9 15:28:55 nmhost01 clusvcmgrd: [5670]: <info> service info: stty: standard input: Inappropriate ioctl for device Jul 9 15:28:55 nmhost01 clusvcmgrd: [5670]: <info> service info: stty: standard input: Inappropriate ioctl for device Jul 9 15:28:56 nmhost01 clusvcmgrd: [5670]: <info> service info: /dev/sde6 is not mounted Jul 9 15:28:58 nmhost01 clusvcmgrd: [5670]: <notice> service notice: Stopped service wms_cb2 ... Now, the "stty" errors are not our problem; they are the result of user-plug-in scripts associated with service. The problems in question are (1) after this happens, when CFG_WriteFile() is called, it appears that the files are no longer renamed/updated for some reason and (2) mount is complaining that /etc/mtab doesn't exist while trying to create the lock file named "/etc/mtab~"!
Created attachment 64864 [details] Log file
I have found some temp files laying around in /etc. As a possible fix, I've implemented file locking (fcntl) and synchronous operation in parseconf.c in my sandbox. Going to do some more testing.
Potential fix in pool. Will have testing performed before closing.
Sent package with potential fix to person who is having this problem. Ideally fixes the temp file problem, but not necessarily the mtab problem.
This has not been reproduced by testers. I'm closing it.