What has happened yesterday evening (around 23:00 utc)?
Both my nodes failed chain part test.
same thing happened to my nodes.
A lot of nodes failed that test at that time, including my nodes.
I will have a closer look at it tomorrow.
I looked into the issue for hours but i cannot figure out what happened. The master just didnāt record any chain part test results for about 2 hours. Then everything went back to normal.
Iām not sure if everything back to normal, my node failing at the last round (Nr. 1402) for the āchain partā piece of test as well.
Itās been 10th or 12th fail of tests in case of my nodes, even if Iām 100% sure, that my nodes running well whole the time. Most of that fails were in āresponsivenessā part, and it was due to the long distance (and some network outages āon the roadā) between my nodes (Central Europe) and measuring points (U.S.).
But itās of course a little bit sad and somewhat unfairly, if someone missed deserve reward, just because of troubles with tests. Somebody suggested for change the rules - in the meaning of send the rewards, if there is just one āsingle and lonelyā fail (as itās more probably problem with the tests than with the supernode, in that case). And I intercede for the same. It would be more fair to all, I think.
It seems it is the same Problem like before. In Round 1402 the chain part failure beginning with https://supernodes.nem.io/history/243639).
Really strange.
I restarted the master, hopefully it helps. I donāt see any reason why the tests suddonly should failā¦
And itās here again ā¦ failed (by āchain partā) at round 1406.
What about the suggestion about change the rules of rewards, (at least) until this measuring problem will be successfully solved?
The problem might be due to the fee fork. I updated master and restarted. If it doesnāt help i have to look deeper into it by debugging with a second master on my local computer.
i would rather not introduce another complication. I need to fix the issue, not code a workaround.
why not just allow 1 error in one round if 3 rounds before that where without a error
that would cover many false positive fails
like a unlucky compute power or bandwith test from a node that is 98% of the time inside the specifications
and yes it would also be to some degree a workaround to cover rare bad chain part results
No workarounds.
I pinned down the problem with the chain part test. Some japanese user created mosaics with kanjis in the description. Unfortunately the master did not like that. I fixed it (hopefully) and will restart the master as soon as round 1412 is over.
ItĀ“s nice to hear that.
15000 Kanjis - no one of the Japanese knows all the Kanjis.
I have a bit āempathyā with the master. ā¦
.
I also had some random fails, but this should not be a problem since every day the same amount is divided between nodes so it will eventually average out.
Still any problems with chain part test?
+1 supernode quality goal must be 100%
I have problems with my supernode. Maybe not enough diskspace.There are a lot of logfiles in the logfolder. Can i delete old logs?
Yes, I do it once a month.
Ok, thanks
@BloodyRookie please check NAFUND-BUKIOS-TMD4BN-XL7ZFE-735QHN-7A3FBS-6CMY, Iād like to change IP of my Fund Supernodes.