[listenbrainz-android] 14dependabot[bot] opened pull request #456 (03dev…dependabot/gradle/dev/androidGradlePlugin-8.5.1): Bump androidGradlePlugin from 8.5.0 to 8.5.1 https://github.com/metabrainz/listenbrainz-andr...
[listenbrainz-android] 14dependabot[bot] opened pull request #457 (03dev…dependabot/gradle/dev/io.socket-socket.io-client-2.1.1): Bump io.socket:socket.io-client from 2.1.0 to 2.1.1 https://github.com/metabrainz/listenbrainz-andr...
[listenbrainz-android] 14dependabot[bot] opened pull request #458 (03dev…dependabot/gradle/dev/org.jsoup-jsoup-1.18.1): Bump org.jsoup:jsoup from 1.17.2 to 1.18.1 https://github.com/metabrainz/listenbrainz-andr...
[listenbrainz-android] 14dependabot[bot] opened pull request #459 (03dev…dependabot/gradle/dev/io.sentry.android.gradle-4.10.0): Bump io.sentry.android.gradle from 4.9.0 to 4.10.0 https://github.com/metabrainz/listenbrainz-andr...
[listenbrainz-android] 14dependabot[bot] opened pull request #460 (03dev…dependabot/gradle/dev/com.google.devtools.ksp-2.0.0-1.0.23): Bump com.google.devtools.ksp from 2.0.0-1.0.22 to 2.0.0-1.0.23 https://github.com/metabrainz/listenbrainz-andr...
Oh i see. I was under the impression they are server auction machines.
zas[m]
to be sure, sparker cluster is Jackson5 machines right? They are all AX41-NVMe
lucifer[m]
Yes indeed. Makes sense
ax42 looks great if we could replace the drives to be at least 1TB.
I see an option to add a 1TB drive but doesn't seem possible to remove the 512gb ones.
zas[m]
What's the purpose of new machines? Increase the capacity of the cluster? or different role?
lucifer[m]
For now load testing a LB solr installation and later yes increasing the capacity of the cluster.
mayhem[m] had totally forgotten that we upgraded them to VMs
zas[m]
lucifer: for load testing, wouldn't virtual machines good enough? That's temporary right?
mayhem: spark cluster are physical machines, AX41-NVMe x 5
lucifer[m]
Yeah a vm would work too but even the 32gb ram one costs same as ax-42
We need one machine to run the solr installation and that should have lots of ram.
The second machine to run jmeter can be a smaller vm.
akshaaatt[m] has quit
akshaaatt[m] joined the channel
zas[m]
lucifer: ok, perhaps we should then go for AX52 (for SOLR) +1 VM for jmeter (no setup fee, temporary use). Then we can upgrade spark nodes to AX52 (increased cpu perf, more disk space, newer hardware), and decide if we keep 5 or 6. What do you think?
hmmm, J5 nodes are using raid0 disks so they already have 1Tb disk space
lucifer[m]
zas: sorry, i just realised that we could pay for hetzner VMs hourly. two vms would do fine. one 64gb one, one 16gb one. dedicated vcpus for both.
ideally we are done with both in a week.
zas[m]
they will run software independently for the rest right? (no consul?)
s/for/from/
lucifer[m]
zas: yes.
just need to be communicate to each other, preferably internal vlan? so that nothing needs to be exposed to public internet.
zas[m]
You need them now?
hmmm, but VMs will not have much disk space, max is 960Gb for ccx63 (dedicated vCPU), and we need to ask for d. vcpu limit increase for that.
zas[m] uploaded an image: (120KiB) < https://matrix.chatbrainz.org/_matrix/media/v3/download/matrix.org/mcGNyxiuXdYOphOYkkILvFeO/image.png >
lucifer[m]
zas: CCX23 and CCX43 would work fine. enough disk space for this test.
zas[m]
ok, tell me when you need them, We need to keep them the shortest time possible to limit costs.
lucifer: this is quite interesting regarding strings. I wonder if some of these insights could help us sometime down the line: https://cedardb.com/blog/german_strings/
[bookbrainz-site] 14MonkeyDo merged pull request #1103 (03import-entities…imports-2024): Make approval of imported entities work & other improvements https://github.com/metabrainz/bookbrainz-site/p...
zas[m]
mayhem: sorry I was afk, I didn't fix them yet, but now it's done ;)