... Your "spike" for the day is roughly 2,000,000 credits above the day before. This relates to 3 FGRPB1 tasks that granted 693,000 instead of 693 credits.
I took quite a while to compose my previous message and was unaware of your further investigation when I finally posted it. I'm very pleased that you have identified the problem as one of granting 693,000 instead of 693. I just had a look on the website at the credit awards for FGRPB1 on both my high RAC hosts and they both show the normal 693 and not the fictitious 693,000. Why don't we see 693,000 or am I missing something??
When you do your next series of validations, I'd prefer it if you would use the normal amount of credit rather than an inflated value. As you can see from the log snippets, I go to a lot of trouble to monitor the health of my hosts and stable RAC is something I like to have. Once disturbed, it does take quite a while to get back to the normal range :-).
I also have no clue why it granted 100 times the normal amount and why I can't find anything about this in the database. This is what puzzles me the most too. Every database field that contains the granted credit for those workunits contains '693'. It's like the value increased shortly before the call to grant credit (and update the host, user, team total_credit) and returned to normal when the result and workunit entries where updated. There is no reason for that in the code and it seems to only happened around 10:18.
I'm sorry this throws off your script but from our perspective the validator should always grant the credit we specify at workunit generation. Those values are all sound and "normal".
I also have no clue why it granted 100 times the normal amount and why I can't find anything about this in the database.
It's actually 1,000 times the normal amount :-).
Quote:
I'm sorry this throws off your script but from our perspective the validator should always grant the credit we specify at workunit generation. Those values are all sound and "normal".
I think we are both slightly misunderstanding each other. I took your previous comment about the next validation ("I'll try to reproduce it") to mean that you were going to try to make each validated task be awarded 693,000 credits. I was trying to indicate that I'd prefer 693 credits which is the the 'normal' amount.
My script is not being "thrown off". It pays no attention to RAC ratios greater than 1 :-). I was trying to say that I'd prefer not to get 693,000 credits for a single task where the standard award specified at workunit generation was just 693. I like to see RAC values that are meaningful. An award of 693,000 for a single task makes the ongoing numbers rather meaningless for quite a while.
I was wondering why my RAC on one host (prodigy_orange) had gone up so much, it seemed to double overnight.
Though it might be to do with some networking issues ive got at the moment, or perhaps a lot of pending tasks had gone though.
Looking though the stats there is a big jump in credits on the 3rd,over 700,000 more than normal.
I couldn't reproduce the 1,000 times credit granting FGRPB1 tasks again. I hope that this will not come up again when I'm not looking. I turned the validator on automatic and if you see a credit spike again please tell me.
RE: RE: ... Your "spike"
)
I also have no clue why it granted 100 times the normal amount and why I can't find anything about this in the database. This is what puzzles me the most too. Every database field that contains the granted credit for those workunits contains '693'. It's like the value increased shortly before the call to grant credit (and update the host, user, team total_credit) and returned to normal when the result and workunit entries where updated. There is no reason for that in the code and it seems to only happened around 10:18.
I'm sorry this throws off your script but from our perspective the validator should always grant the credit we specify at workunit generation. Those values are all sound and "normal".
RE: I also have no clue why
)
It's actually 1,000 times the normal amount :-).
I think we are both slightly misunderstanding each other. I took your previous comment about the next validation ("I'll try to reproduce it") to mean that you were going to try to make each validated task be awarded 693,000 credits. I was trying to indicate that I'd prefer 693 credits which is the the 'normal' amount.
My script is not being "thrown off". It pays no attention to RAC ratios greater than 1 :-). I was trying to say that I'd prefer not to get 693,000 credits for a single task where the standard award specified at workunit generation was just 693. I like to see RAC values that are meaningful. An award of 693,000 for a single task makes the ongoing numbers rather meaningless for quite a while.
Cheers,
Gary.
I was wondering why my RAC on
)
I was wondering why my RAC on one host (prodigy_orange) had gone up so much, it seemed to double overnight.
Though it might be to do with some networking issues ive got at the moment, or perhaps a lot of pending tasks had gone though.
Looking though the stats there is a big jump in credits on the 3rd,over 700,000 more than normal.
Seems I got one of bonus tasks.
I couldn't reproduce the
)
I couldn't reproduce the 1,000 times credit granting FGRPB1 tasks again. I hope that this will not come up again when I'm not looking. I turned the validator on automatic and if you see a credit spike again please tell me.
RE: Everyone else gets a
)
Please tell me your anti-spike was not -666
RE: Please tell me your
)
No it was not, and no daemons were harmed in the process.