Search results

Hi thought so, thx. ("no compatibility guarantee for clusters with v5.x and v7.x nodes" <- good point) it's not my first dist-upgrade of clusters, but this one is a very time- (and coordination-) consuming 'cause not all VMs can live-migrate (5x doesnt allow live-migration with local storage...

Hi I have to dist-upgrade a cluster currently running 5.x (already corosync-3). your recommendations are welcome: can I dare to dist-upgrade each host 5->6->7 (not 5->7 but 2 dist-upgrades while the rest of the cluster ist still 5.x)? or should I better go the slow way and dist-up to 6 the...

this depends on the cabling of controller/motherboard and backplane. if you assume slot 1 is most up/left 9f your slots. thats ok. mostly. but what is slot 2? the disk right of it or below it? had both. and I had machines where slot 1 is the most left/down(bottom) slot. so no pci or any other...

Beides. Einen VLAN-Switch bekommst du fuer ein paar Euro (kurz bei geizhals reingeschaut, ab 20 EUR 5 ports). Ich würde einen gebrauchten Server in Betracht ziehen (zb Intel XEON E3 oder E5) wenn er nicht durchlaufen muss/soll oder einen alten Laptop. Mein Test-Cluster besteht zb aus Dell...

ich würde mehr wert auf möglichst viel ram und schnelle cpu legen. das ist imo wichtiger um wirklich testen zu können. mit vlans kannst du im testbetrieb alles an einer nic machen, zuwenig ram oder langsame cpu behindern aber wirklich.

normally systemd should recognize remote mounts automatically, but you can try _netdev as mount option in fstab. this tells systemd to mount after network is online.

I dont know if zfsunlock has a feature to do that, but on the shell you can accomplish this with "expect" regards Peter

we too would appreciate neutral language

thx. the destination is a fully licenced cluster anyway. the open question was "when" to do the licensing of the new cluster hosts (that will stay while the oldest ones will drop out).

we are looking forward to the same process. additional "problem" is, that we wont migrate all hosts of the existing cluster and retire some of the older hosts. so fo some time, we would have a mixed cluster licensed/unlicensed. still unsure what way to go. hints/tips?

i'd take a look at ansible, puppet, chef etc... we are using ansible for all deployments. as you are speaking of a development environment, maybe there are other deployment mechanisms (github actions and such things) to distribute your certs too. if you use letsencrypt you can automate the...

maybe tailscale (with headscale) and it's namespaces would fit for your needs

^ md1 as I see in the screenshot is root-fs

maybe tune2fs -m0 /dev/md0 frees some bytes for you.... to keep the system working and clean up

I'd try this in first place.

is it stuck or counting? when it's counting just wait till finished. if it's stuck, do a RAM test, maybe the änew RAM has defects.

you dont define the usecase very well but maybe an asynchronous solution would fit. i'd go for nextcloud and tailscale (why expose pve unnecessarily)

you need stackable switches if you want to use active/active and redundancy

what would be the advantage to raid-1?

maybe power consumption is an argument pro ssd. with 2 x ssd you could substitute 5 spinning disks and get more performance.