| Time |
Nick |
Message |
| 07:09 |
|
collum joined #evergreen |
| 07:13 |
JBoyer |
Sorry, I lost track of the conmversation yesterday. I think that turning caching off entirely for /eg/opac/results would "fix it" but I also think it's a bug that Chromium is caching a 302 result (caching a 301 might be defensible). |
| 07:16 |
JBoyer |
That it's a bug is even more likely because Chromium is doing that and Chrome is not. Firefox (139 on Pop_OS! / Ubuntu) also just bounced once as expected for me. |
| 07:20 |
JBoyer |
I would expect that the presence of a query string pretty should limit caching, but I know I'm behind the times on The Way Things Are Done in browsers. |
| 08:29 |
|
dguarrac joined #evergreen |
| 09:18 |
|
sandbergja joined #evergreen |
| 09:30 |
Bmagic |
csharp_: Quipu is still an issue then? |
| 09:46 |
|
Guest53 joined #evergreen |
| 09:49 |
csharp_ |
Bmagic: I'm about to be off for a week and a half, so I just disabled the apache var for now - I'll look again when I'm back on June 30 |
| 09:49 |
Bmagic |
coolio |
| 09:49 |
csharp_ |
I have been using nginx to redirect specific user agents to a 403 |
| 09:50 |
Bmagic |
gotcha |
| 09:50 |
csharp_ |
that and geo blocking + some specific IP ranges blocked, etc. |
| 09:50 |
csharp_ |
it's a war of attrition |
| 09:50 |
Bmagic |
yep, same |
| 09:51 |
Bmagic |
We had Anubis on production for a day and the results were amazing |
| 09:51 |
Bmagic |
100% -> 1% in less than 3 minutes |
| 09:51 |
csharp_ |
wow |
| 09:56 |
|
Jaysal joined #evergreen |
| 10:03 |
|
Jaysal joined #evergreen |
| 10:03 |
Jaysal |
testing |
| 10:24 |
csharp_ |
Jaysal: it worked! |
| 10:34 |
|
Dyrcona joined #evergreen |
| 10:44 |
|
sandbergja joined #evergreen |
| 11:09 |
|
Christineb joined #evergreen |
| 11:36 |
|
jihpringle joined #evergreen |
| 11:53 |
pinesol |
News from commits: Release notes for 3.15.2 <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=5e27775048ae4864c0afc594ef319656dd77f5e7> |
| 11:53 |
pinesol |
News from commits: Release notes for 3.14.7 <https://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=aba82d6c276ecd807883392640c23e670c1e5d27> |
| 12:24 |
jeffdavis |
I mentioned this yesterday but with bug 2113979 I'm concerned about redirect loops spamming the server when cookies are disabled, is there an obvious way to prevent that? |
| 12:24 |
pinesol |
Launchpad bug 2113979 in Evergreen 3.15 "Put a hurdle in front of basic scraping" [Medium,New] https://launchpad.net/bugs/2113979 |
| 12:36 |
|
jihpringle joined #evergreen |
| 12:50 |
|
jihpringle joined #evergreen |
| 13:35 |
|
ChurchillComputi joined #evergreen |
| 13:35 |
|
kmlussier joined #evergreen |
| 14:02 |
|
jihpringle joined #evergreen |
| 14:06 |
jeffdavis |
Bmagic: are you able to share any details about how you set up Anubis? |
| 14:57 |
Bmagic |
jeffdavis: sure! https://github.com/mcoia/eg-docker/commit/d36a3b295b41613032d381aa5a3d15422efd6c11 |
| 15:11 |
justdoglet |
If you're struggling with interrupt load on your LB/firewall system and have a lot of iptables rules, switch your DROPs to an ipset. Makes a world of difference. |
| 15:12 |
justdoglet |
(And install ipset-persistent along with netfilter-persistent and iptables-persistent if you use those, so the ipsets can be saved and reloaded at boot.) |
| 15:12 |
|
kmlussier1 joined #evergreen |
| 15:43 |
csharp_ |
seeing an issue I've not noticed before where a client repeatedly makes this call, over and over, many times per second: "open-ils.pcrud open-ils.pcrud.search.pgt null, {"parent":null}, {"flesh":-1,"flesh_field |
| 15:43 |
csharp_ |
s":{"pgt":["children"]}}" |
| 15:44 |
csharp_ |
it's contributing to high system load on our app servers |
| 15:45 |
csharp_ |
if the bots weren't enough |
| 15:54 |
Bmagic |
hmmm |
| 16:07 |
jeffdavis |
csharp_: http://irc.evergreen-ils.org/evergreen/2019-07-11#i_411837 |
| 16:09 |
jeffdavis |
temporary account operator change timeout getting stuck in a loop and spamming in a way we've seen the client do elsewhere maybe? |
| 16:10 |
jeffdavis |
there's supposed to be proper handling for null authtokens in pcrud these days though I thought |
| 16:11 |
jeffdavis |
(also belated thanks for that Anubis info Bmagic!) |
| 16:13 |
Bmagic |
jeffdavis: YW. Also: that repo has 3 different builds, each with basically the same changes to support Anubis. The one that has it enabled by default is "generic-dockerhub". Turning it on/off is a matter of including a URL to download the debian package or not. |
| 16:14 |
csharp_ |
jeffdavis++ # thanks |
| 16:14 |
|
jihpringle joined #evergreen |
| 16:16 |
csharp_ |
ah I think this from JBoyer is the key: "I think it's temporary (5 min) account operator changes timing out" |
| 16:26 |
csharp_ |
does seem like bug 1990306 should have fixed this - maybe we missed one? |
| 16:26 |
pinesol |
Launchpad bug 1990306 in Evergreen 3.9 "Flood of open-ils.actor.user.has_work_perm_at.batch requests for VIEW_USER with null authtoken" [High,Fix released] https://launchpad.net/bugs/1990306 |
| 16:28 |
|
kworstell-isl joined #evergreen |
| 16:42 |
|
kmlussier joined #evergreen |
| 16:51 |
|
kmlussier left #evergreen |
| 17:09 |
|
ChurchillComputi joined #evergreen |
| 18:42 |
|
jihpringle joined #evergreen |
| 18:49 |
|
jihpringle joined #evergreen |
| 19:08 |
|
jihpringle joined #evergreen |
| 19:33 |
|
jihpringle joined #evergreen |
