Full record view with mra fleshing causes an internal timeout
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Evergreen |
New
|
Undecided
|
Unassigned |
Bug Description
Evergreen 2.12.9 (fork)
OpenSRF 2.5.2
Ubuntu Xenial 16.04 lts
PostGresql 9.5
This week we upgraded from 2.7.1 to 2.12.9.
For records that share sometimes thousands of similar titles (which is typical for our visual record descriptions) our staff noticed a timeout for a full view with the staff client and web browser.
The view is initiated with the WWW Record.pm module, where information on relates records is gathered, as shown here in red at line 92 and 228:
https:/
In the backend this translates into a request-response similar to this srfsh CLI:
request open-ils.cstore open-ils.
Received no data from server
-------
Request Time in seconds: 16.894166
-------
The log:
srfsh 2018-02-21 15:12:01 [DEBG:10483:
srfsh 2018-02-21 15:12:01 [DEBG:10483:
srfsh 2018-02-21 15:12:01 [DEBG:10483:
srfsh 2018-02-21 15:12:01 [DEBG:10483:
srfsh 2018-02-21 15:12:01 [DEBG:10483:
srfsh 2018-02-21 15:12:01 [INFO:10483:
srfsh 2018-02-21 15:12:01 [DEBG:10483:
srfsh 2018-02-21 15:12:01 [INFO:10483:
srfsh 2018-02-21 15:12:01 [DEBG:10483:
Whereby the apache web requests would wait apparently forever for a response it will never receive.
To work around this, I removed the mra field from the call.
We expect the list with the mra would either be limited if too large... or possibly disabled with a configuration option.
tags: | added: perl |