In addition to the one hit isbn searches and metarecord searches where we have see the 70 second delay, I have also seen the same delay in basic keyword searching.
I conducted a series of two-word keyword searches on a 2.12.2 training server with production data while logged in as a patron with pref_ou=SAL. Out of 26 searches, the 11th, 14th, and 24th failed with a timeout at ~70 seconds. I performed the failed searches later in the day and they returned hits normally.
A summary of the searches:
8:48am - apple kick - no hits (timeout)
8:55am - apple nest - no hits (timeout)
9:15am - apple water - no hits (timeout)
Successful re-searches:
9:18am - apple kick - succeeded, 15 hits
9:23am - apple nest - succeeded, 27 hits
9:35am - apple water - succeeded, 141 hits
I have captured timelogs and osrfsys logs for these searches and uploaded the files here:
In addition to the one hit isbn searches and metarecord searches where we have see the 70 second delay, I have also seen the same delay in basic keyword searching.
I conducted a series of two-word keyword searches on a 2.12.2 training server with production data while logged in as a patron with pref_ou=SAL. Out of 26 searches, the 11th, 14th, and 24th failed with a timeout at ~70 seconds. I performed the failed searches later in the day and they returned hits normally.
A summary of the searches:
8:48am - apple kick - no hits (timeout)
8:55am - apple nest - no hits (timeout)
9:15am - apple water - no hits (timeout)
Successful re-searches:
9:18am - apple kick - succeeded, 15 hits
9:23am - apple nest - succeeded, 27 hits
9:35am - apple water - succeeded, 141 hits
I have captured timelogs and osrfsys logs for these searches and uploaded the files here:
https:/ /drive. google. com/drive/ folders/ 0B-CjwZFtLEK6VG dNX241RXpmaXM? usp=sharing
The failed searches all seem to have a "flushing data from socket" entry prior to the 70 second delay.