Tired. Exhausted. Smelly
Ended my last primary on-call shift of the year just now, and not a minute too soon. I like to say that my job is at the intersection of system administration and adrenaline because for 37 hours a weekend we are constantly solving issues where data loss is not an option and speed is everything.
Anyway, two choice quotes from my weekend, emphasis mine:
From a Sybase ASE error log:
2008/12/07 02:34:34.40 server Error: 9970, Severity: 20, State: 1 2008/12/07 02:34:34.40 server DBCC cannot update the finish time in dbcc_operation_log table for this operation(opid = ’182′) of database ‘sybsystemprocs’. This can be patched by executing sp_dbcc_patch_finishtime.
I love software that has been patched to tell you how to actually fix the issue but won’t actually fix it for you.
Next up, an email from a first-level system administrator:
We see that the host is now responding. But host is on high load. Please check your applications. piias1332 /ms/user/v/venusrin 63$ rup pisdb38 pisdb38 up 29 days, 2:42, load average: 1.77 1.73 1.62
Um. No. Just no.
2 Comments »
RSS feed for comments on this post. TrackBack URI
why no?
Load average on Solaris is defined as the number of running processes plus the number of runnable processes. (On Linux, it includes the number of outstanding I/O requests). A load average of 1 means there is one process that can be run for the duration of the sample. Having one process around is not a high load.