| Time |
Nick |
Message |
| 04:31 |
|
jonadab joined #evergreen |
| 05:24 |
|
byran joined #evergreen |
| 05:25 |
byran |
help |
| 05:25 |
byran |
hello |
| 05:25 |
byran |
i ama newbee |
| 05:26 |
byran |
and have installed a local version of evergreen and would like help in migrating from my legacy system if anyone can help |
| 11:15 |
dbs |
a single apache process at 2.9% of 16GB of RAM - that's bloat (with 1000 maxconnsperchild, too)! http://pastebin.ca/3719227 |
| 13:05 |
|
bmills joined #evergreen |
| 13:08 |
|
bmills1 joined #evergreen |
| 13:08 |
|
ssieb joined #evergreen |
| 13:32 |
* dbs |
wonders if 'DBD::Pg::st execute failed: ERROR: syntax e |
| 13:32 |
dbs |
rror at or near "["#012LINE 2: WITH x9122048_author_personal[Wittgenstein, Ludwig 1889-1951' is a sign of needing PG > 9.1 in 2.10 |
| 13:33 |
dbs |
that's pretty weird-looking CTE to these eyes, though, I wouldn't expect SQL to suddenly start allowing '[' as a char in identifiers |
| 13:34 |
dbs |
(that's part of 'WITH x9122048_author_personal[Wittgenstein, Ludwig 1889-1951 subject_xq AS (SELECT ...' for context) |
| 13:36 |
dbs |
769 errors of that sort in the last 12 hours, that might be doing a number on open-ils.storage |
| 13:43 |
* dbs |
peers at QueryParser::query_plan::facet->table_alias |
| 14:02 |
dbs |
hmm, calling table_alias() on the result of do{QueryParser::query_plan::facet}->new(name => 'author_personal[Wittgenstein, Ludwig 1889-1951 subject'); gives me that |
| 14:03 |
dbs |
perhaps the result of web crawlers generating facets in unexpected combinations? |
| 14:04 |
dbs |
ah there we go: EGWeb: [bre search] author|personal[Wittgenstein, Ludwig |
| 14:04 |
dbs |
1889-1951 subject:Logic depth(0) |
| 14:05 |
dbs |
looks like QP is barfing over that |
| 14:11 |
dbs |
no idea why it's missing the closing ] for the facet. I can't reproduce after a few minutes of trying to click through results :/ |
| 15:08 |
|
bmills joined #evergreen |
| 15:35 |
miker |
dbs: yeah, QP doesn't see a facet there because of the missing ], instead it sees a field called "personal[Witt....subject" because subject is filling a : |
| 15:38 |
miker |
fields aren't registered with QP, iirc, just classes. we could tighten the field part of the regex to, say, \w+, maybe... |
| 15:38 |
miker |
um, followed by, not filling. on phone, autocorrect :) |
| 17:28 |
dbs |
jeepers. bloatiest apache processes are now at 560992/554696/502832 RES memory, ~ 3:00 TIME in top. > 3% of 16GB RAM each! |
| 17:29 |
dbs |
maybe I'll ramp the maxchildconnections down to 500 or something. that's nuts. |
| 17:48 |
dbs |
interesting, simple2zoom seems to go nuts occasionally attempting to retrieve a record, seems like if it failes to process a record at Supercat.pm:2087 it tries over and over and over again |
| 17:48 |
dbs |
might be the client, too |
| 17:50 |
dbs |
hah, wonderful, looks like it's OCLC hammering our z39.50 server |
| 18:03 |
dbs |
Well, OCLC isn't hammering us. It's just requesting a single record that has Unicode characters, and yaz 4.2.30 / simple2zoom is going nuts |
| 18:12 |
dbs |
... and it doesn't matter if it's yaz 4.2.30 or latest yaz 5. hmm |
| 18:13 |
* dbs |
looks at the MARC for https://laurentian.concat.ca/eg/opac/record/817034?expand=marchtml#marchtml and spots the 020 subfield null. Well then. |
