OFFICIAL "SERVER PROBLEMS" THREAD

[Madpoo]

Originally Posted by James Heinrich

@Aaron: Is there any reason there's no ORDER BY `exponent` when fetching from the database? Or if that adds appreciable overhead to the query for some reason then it should be easy to sort by exponent in PHP between fetching it from the database and displaying as HTML.

There's no reason we can't add that, and that was going to be my next idea, just to have an "order by exponent" in there.

Otherwise I think it's probably going to fetch them in their assigned order since that's probably the order they're added to the DB (makes sense), but even then who knows...it could be by their assignment ID in which case it'd be totally random.

If you want to add that, I'm cool with it. I'll add it to my mocked up page that has the expirations and stuff, just so I can try that out.

EDIT: Never mind, I went ahead and added it. I noticed on my mock page I'd already added an "order by" clause, so apparently I'd already noticed it at the time I was working on that, and then forgot about it.

[ric]

Originally Posted by Madpoo

EDIT: Never mind, I went ahead and added it.

Works (again) like a charm. Thanks!

Still curious as to the root cause for that strange behaviour, however.

[garo]

I like the new workload page!

Just wondering how practical it would be to display the current category of a machine on the CPU page - i.e. will it get Cat 0/1/2/3/4 LL/DC based on current settings. In fact it might be even better to list the criteria such as #days work, expired exponents, expected days to complete etc. so that we can identify what is stopping a machine from getting an exponent of a preferred category.

Case in point: my main machine chanakya suddenly started receiving Cat 2 exponetns a month ago. I cannot see any expired or triple-checked exponents in the last 120 days, it completes a DC every 14 days or so. Days to complete is 5 and yet... I've been scratching my head for the past while and it would be good to have that information on the CPU page.

[petrw1]

Cores 1 and 2 are good.
For some reason on cores 3 and 4 it gives 10 days as the completion days (about right) for the 2 waiting assignments ... but 120 and 119 days for the active assignments (should be about 3 days)

Rocky	1	38,252,147 	D	LL, 53.70%	7	2016-11-16	2016-11-23	2016-11-24	2016-11-26	3
Rocky	1	38,286,673 	D	               	0	2016-11-23	2016-11-23	2016-11-24	2016-12-03	10
Rocky	2	38,252,161 	D	LL, 50.40%	7	2016-11-16	2016-11-23	2016-11-24	2016-11-26	3
Rocky	2	38,287,969 	D                       0	2016-11-23	2016-11-23	2016-11-24	2016-12-03	10
Rocky	3	38,286,719 	D                       0	2016-11-23	2016-11-23	2016-11-24	2016-12-03	10
Rocky	3	38,250,307 	D	LL, 53.70%	8	2016-11-16	2016-11-23	2016-11-24	2017-03-23	120
Rocky	4	38,286,893 	D	        	0	2016-11-23	2016-11-23	2016-11-24	2016-12-03	10
Rocky	4	38,245,429 	D	LL, 54.20%	8	2016-11-15	2016-11-23	2016-11-24	2017-03-22	119

[snme2pm1]

Originally Posted by petrw1

120 and 119 days for the active assignments

One would imagine that your prime95 periodic completion message to PrimeNet indicated 2017-03 completion for two of those.
Is that evident in your logs?

[petrw1]

[Comm thread Nov 22 23:56] Updating computer information on the server
[Comm thread Nov 22 23:56] Exchanging program options with server
[Comm thread Nov 22 23:56] Exchanging program options with server
[Comm thread Nov 22 23:56] Exchanging program options with server
[Comm thread Nov 22 23:56] Exchanging program options with server
[Comm thread Nov 22 23:56] Exchanging program options with server
[Main thread Nov 22 23:56] Restarting all worker threads.
[Comm thread Nov 22 23:56] Sending expected completion date for M38252147: Nov 26 2016
[Comm thread Nov 22 23:56] Sending expected completion date for M38252161: Nov 26 2016
[Main thread Nov 22 23:56] Restarting all worker threads using new settings.
[Comm thread Nov 22 23:56] Sending expected completion date for M38250307: Mar 23 2017
[Comm thread Nov 22 23:56] Sending expected completion date for M38245429: Mar 22 2017
[Comm thread Nov 22 23:56] Done communicating with server.
[Comm thread Nov 23 02:13] Exchanging program options with server
[Comm thread Nov 23 02:13] Done communicating with server.

Then I simply submitted new end dates again

[Comm thread Nov 23 17:13] Updating computer information on the server
[Comm thread Nov 23 17:13] Sending expected completion date for M38252147: Nov 26 2016
[Comm thread Nov 23 17:13] Sending expected completion date for M38286673: Dec 03 2016
[Comm thread Nov 23 17:14] Sending expected completion date for M38252161: Nov 26 2016
[Comm thread Nov 23 17:15] Sending expected completion date for M38287969: Dec 03 2016
[Comm thread Nov 23 17:16] Sending expected completion date for M38250307: Nov 26 2016
[Comm thread Nov 23 17:18] Sending expected completion date for M38286719: Dec 03 2016
[Comm thread Nov 23 17:19] Sending expected completion date for M38245429: Nov 26 2016
[Comm thread Nov 23 17:19] Sending expected completion date for M38286893: Dec 03 2016
[Comm thread Nov 23 17:19] Done communicating with server

[Prime95]

Warning: mersenne.org may be offline some during the next few days.

Our ISP sent us an automated warning that we had triggered too much data traffic to a specific target over a period of time. In other words, we looked like we were part of a ddos attack. When this happened three times, we were knocked offline for 45 minutes.

Unfortunately, madpoo is unavailable until Sunday to do a deep dive into our logs, run anti-virus scans, etc. So if it happens again, we may suffer more 45 minute outages.

I looked at the logs briefly, did a whois on the target, and learned little.

[NBtarheel_33]

Originally Posted by Prime95

Warning: mersenne.org may be offline some during the next few days.

Our ISP sent us an automated warning that we had triggered too much data traffic to a specific target over a period of time. In other words, we looked like we were part of a ddos attack. When this happened three times, we were knocked offline for 45 minutes.

Unfortunately, madpoo is unavailable until Sunday to do a deep dive into our logs, run anti-virus scans, etc. So if it happens again, we may suffer more 45 minute outages.

I looked at the logs briefly, did a whois on the target, and learned little.

Come on now, George, surely you know that it's Putin . We recently double-checked the 45th Mersenne prime, Trump is the 45th US president, the pieces all fit together. Just wait until those cheeky Russians hack Prime95 and cause every LL residue to return 1917191719171917.

[garo]

Originally Posted by NBtarheel_33

Come on now, George, surely you know that it's Putin . We recently double-checked the 45th Mersenne prime, Trump is the 45th US president, the pieces all fit together. Just wait until those cheeky Russians hack Prime95 and cause every LL residue to return 1917191719171917.

LOLOLOLOL! Super post.

[Madpoo]

Originally Posted by Prime95

Warning: mersenne.org may be offline some during the next few days.

Our ISP sent us an automated warning that we had triggered too much data traffic to a specific target over a period of time. In other words, we looked like we were part of a ddos attack. When this happened three times, we were knocked offline for 45 minutes.

Unfortunately, madpoo is unavailable until Sunday to do a deep dive into our logs, run anti-virus scans, etc. So if it happens again, we may suffer more 45 minute outages.

I looked at the logs briefly, did a whois on the target, and learned little.

I did finally look into it. The bulk of the traffic came from some IP in Switzerland, of all places.

Unfortunately the colocation provider only said it was TCP traffic, but no port or anything.

Getting blackholed as a target of a DDOS definitely sucks (if that's what it was... I wasn't so sure based on the info they sent). My day job had that happen a few weeks back where, for whatever reason, a DDOS hit our primary website... absolutely no reason whatsoever. I just imagine some script kiddie in mommy's basement saying something about "I did it 4 the lulz".

EDIT: In case you didn't know, an ISP will frequently blackhole (null route) the target of a DDOS in order to protect the rest of their customers. I don't blame them for that, Akamai even had to do that after that *massive* DDOS recently. It just sucks.

[LaurV]

Originally Posted by Prime95

I looked at the logs briefly, did a whois on the target, and learned little.

Originally Posted by NBtarheel_33

Come on now, George, surely you know that it's Putin

(my bolding)

Hahaha, this is even funnier, extremely funny, in Romanian.. (google translation link).
Can't stop laughing, the colleagues in the office are looking to me like "hmmm, the farang got a diode burned in his brain"....