Troubleshooting
Problem
It is possible for a directory in a directory storage pool to enter a state where read operations are successful but all writes to containers in that directory/filesystem fail?
Symptom
Inspection of the dsmffdc.log shows the following messages when write operations fail.
[09-20-2018 02:07:51.810][ FFDC_GENERAL_SERVER_ERROR ]: (psfile.c:3190) Error 52 after writing 0 of 4096 bytes. file=/tsm/69/00000000000069d7.dcf, buffer=165ae2000, size=4096. Check prior seek val as well
[09-20-2018 02:07:52.840][ FFDC_GENERAL_SERVER_ERROR ]: (psfile.c:3190) Error 52 after writing 0 of 4096 bytes. file=/tsm/69/00000000000069d7.dcf, buffer=165ae2000, size=4096. Check prior seek val as well
[09-20-2018 02:07:58.497][ FFDC_GENERAL_SERVER_ERROR ]: (psfile.c:3190) Error 52 after writing 0 of 4096 bytes. file=/tsm/69/00000000000069d7.dcf, buffer=165ae2000, size=4096. Check prior seek val as well
Cause
Error 52 in this case directly maps to AIX/Linux errno 52 STALE FILESYSTEM. This condition can result when a SAN or iSCSI attached block device that serves as the backend storage for the failing directory/filesystem disconnects from the host briefly.
Resolving The Problem
An attempt can be made to umount the filesystem in question and then mount it again. However, this often times will fail with 'filesystem busy' due to queued kernel I/O requests to the filesystem. A restart of the OS generally is needed to resolve the issue at that time.
Historical Number
TS001395072
Product Synonym
TSM, Spectrum Protect
Was this topic helpful?
Document Information
Modified date:
12 November 2018
UID
ibm10732970