I’m asking as it causes too much of a load, downloading the metadata from the main server.
Could it be possible it’s caused by a lot of containers?
If it’s not cruical I would shut it down for now.
Same here. On our cluster Metadata is using at times up easily 300mb and sometimes even 600mb of memory, before jumping back down to 100mb and starting to increase again.
I have a some 2gb machines in the cluster, and the behaviour makes Rancher barely usable for that and smaller memory config machines.
Seems like the memory requirements for hosts went up significantly in the upgrade to 1.2+. I’m hoping it’s temporary.
Do you also have a problem that those services cause a high load on the servers? Almost 300%+ CPU load on them, which translates to a load of 22ish on the server?
no, only if the memory usage causes resource contention, an things start failing anyway. otherwise cpu usage keeps low.
Issue is different then. End result of wanting to turn the thing off is the same.
Technically “no”. But it is required by all the other infrastructure services, so yes, unless you want no DNS, healthchecks, or networking.