Wrong Load information in v11

[dasfliege]

We've updated several of our B&R installations to v11. All of them do have the same problem.
The Load/Bottleneck analysis is always showing 0% load on the proxy. That applies to VMWare as well as to Hyper-V integrations.

Do you guys have the same problem? Is Veeam aware of this issue?

[HannesK]

Hello,
might be an issue... do you have a case number? In my environment, the source is really that slow, that a proxy value at 0% is realistic, but every value at 0% sounds strange.

Best regards,
Hannes

[dasfliege]

No, i haven't opened a case yet, as it's "just" a cosmetic issue. I thought that veeam surely is already aware of that. But if that is not the case, i will report it.

[HannesK]

At least not "well known". If you could refer in your case to case number #04709073, that would help.

[dasfliege]

Opened case #04709286

[dasfliege]

Is there someone with v11 who is NOT seeing this problem? We have it on every v11 installation and it still seems like veeam isn't really aware that it seems to be a general v11 problem.

[veremin]

We haven't seen anything like that reported before, and with the current number of V11 installations I'd expect to see at least several similar posts by now. So I recommend continuing investigation with our support team to see if it is possibly some environment-specific issue. Thanks!

[tsightler]

Just to be sure I took a quick glance at the stats on my jobs in the lab and, so far, the bottleneck stats look correct on everything, at roughly what I would expect. Will be interesting to see what support find.

[SteveR]

I'm seeing the issue on 3 different installations. The proxy has reported 0% on every run since they were upgraded. Prior to the upgrade, it was anywhere between 40-50%. This is on both backup and replication jobs for VMware. The source, network, and target percentages have all increased.

[Egor Yakovlev]

Hi Steve,

Do you have support case opened as well?
Please share ID in this thread or send me via PM.

/Thanks!

[Egor Yakovlev]

Update.

Symptoms: Backup\Replication jobs display 0% Proxy load under bottleneck statistics.
Cause: Internal issue, wrong number calculation formula. Issue ID 303112.
Status: Will be fixed in the following update.

/Thanks!

[tsightler]

Seems to mostly impact Windows proxies, which is why I don't see the behavior in my lab, as my lab has only Linux proxies for normal use. A quick test with a Windows based proxy shows the same behavior reported here.

[Gostev]

wrong number calculation formula

Actually, there's no "calculation formula" for this particular counter, because it merely shows the proxy CPU load. Are you sure there's no misunderstanding? May be double-check with R&D to make sure everyone is on the same page... just in case

[Egor Yakovlev]

True, I have picked slightly wrong wording here.
It was a change in the logic that retrieved said number, causing us to get 0% all the time.

[dasfliege]

I'm still working with veeam support on that case. Not because of the display-thing, as this has already been confirmed to be an issue, but because jobs are generally slower since updating to v11.
I've checked that in three different installations and all of the backupjobs generally takes 30-100% longer after updating to v11 than they did before.
Is anyone having the "dislplay-issue" also observing slower backupjob performance since v11?

[HannesK]

I have seen a situation yesterday, where UseUnbufferedAccess = 0 fixed performance issues.

That key is not a general solution or recommendation. It seems to depend on the storage technology used.

As the case is already pretty long... can you maybe tell us what kind of repository / storage you are using?

[dasfliege]

It's not storage dependent. I see the exact same issues in every installation where we updated from v10 to v11. Proxy load is always 0% and jobs do take generally 30-100% longer then before the update.
But to anwer your question, we observed the issue on ReFS repositories lying on StorageSpace JBOD or Synology as well as on Linux repository lying on Nimble Storage.

[HannesK]

the proxy issue is confirmed already as UI issue.

My post was about

all of the backupjobs generally takes 30-100% longer after updating to v11 than they did before.

Do you have a chance to test the key? You can ask support to cross-check with case #04741000 - but I'm fairly optimistic from what you say that you use as hardware.

[dasfliege]

Sure. I will give it a try.

[dasfliege]

Unfortunately that didn't help. Performance is still not at the level it was with v10

[HannesK]

hmm okay - then I only have the block alignment left mentioned in a StoreOnce performance issue thread post410249.html#p410249

Note: as only support analyzes the logs, I only do "educated guessing".