| 22 Sep 2021 |
 cde | but we can check the web server? | 14:44:57 |
 jochensp | I think Hans has access | 14:45:21 |
| cde | _hc: ^ any ideas? tl;dr from the 2 caching and 1 web server, apache 2.4.25 fast, apache 2.4.38 slow. | 14:45:25 |
| jochensp | ..and staging.f-droid.org is .38 | 14:46:12 |
| jochensp | now .25 was slow as well for me | 14:47:47 |
 tepozoa | cde: apologies 2hr engineering zoom, $work stuff | 14:48:00 |
| tepozoa | we're talking about kubernetes, would much rather talk about f-droid :) | 14:49:53 |
| cde | fdroid-k8s wen? :) | 14:50:25 |
 εΉΈη« (πππΎπ/πππΎπ) | I see 148.251.140.42 -> Apache/2.4.25 (0.1-5s) & 149.202.95.241 -> Apache/2.4.38 (5-20s) here (using for x in $(seq 10); do time curl -svI https://f-droid.org 2>&1 | grep -E 'Connected|Apache'; done) and it seems to switch between the two exactly every time in my limited testing. | 15:35:00 |
![@_oftc_Licaon_Kter[xmpp]:matrix.org](./avatar/Licaon_Kter%5Bxmpp%5D) Licaon_Kter[xmpp] | Caching servers....right? | 15:38:40 |
| tepozoa | hey off my zoom, I can test anything. I can even spin up some VMs in any datacenter we want to use to test, I just happen to have some servers alive in Dallas and Toronto. what can I do that would help out here? | 15:40:49 |
| jochensp | tepozoa: sadly not much, the issue is known and out of our control (besides there is ongoing work on moving stuff to a CDN) | 15:41:33 |
 _hc | someone needs to take that over, not worth much troubleshooting the existing setup. There are two prototypes for doing it | 15:42:32 |
| cde | _hc: would you mind doing a quick apache version check on the server you do have access to? | 15:42:53 |
| _hc | i have access to none of the front end caches | 15:43:15 |
| cde | right, but the web server?
We're getting the version from the header Via | 15:43:53 |
| tepozoa | In reply to @cdesai:matrix.org
two different apache versions reported in both
First: Apache 2.4.25
Second: Apache 2.4.38 these are so similar, I would suspect that it's a configuration issue (where I would start debugging first) | 15:44:24 |
| tepozoa | I've upgraded a metric f-ton of apache in my career, rarely does a version upgrade result in negative network performance of this measurable amount | 15:45:16 |
| tepozoa | * I've upgraded a metric f-ton of apache in my career, rarely does a version upgrade result in negative network performance of this measurable amount | 15:45:24 |
| jochensp | tepozoa: we don't even know the configuration of the caching servers | 15:45:54 |
| tepozoa | nod :( | 15:46:15 |
| tepozoa | is f-droid using a hosted service, that why? | 15:46:46 |
| jochensp | nop, just that the F-Droid founder is doing it | 15:47:22 |
| _hc | and its something we're looking for a new maintainer for | 15:51:18 |
| tepozoa | I know there's been talk of cloudflare and it's problems for some people (captchas on tor, etc.) - has anyone f-droid official approached KeyCDN? https://www.keycdn.com/open-source-cdn | 15:59:04 |
 deimos | https://gitlab.com/fdroid/admin/-/issues/229 | 16:22:17 |
 proletarius101 | In reply to @tepozoa:matrix.org
I know there's been talk of cloudflare and it's problems for some people (captchas on tor, etc.) - has anyone f-droid official approached KeyCDN? https://www.keycdn.com/open-source-cdn Key CDN retains IP address logs for an non-promised period of time | 16:38:25 |
| proletarius101 | But actually i haven't directly contacted them | 17:12:10 |
| proletarius101 | you may have a try. the open source account needs you to contact them directly anyway. maybe you can also ask their log retention policy | 17:13:03 |
| proletarius101 | tepozoa: | 17:13:13 |
