Server outrages - Page 4

mdettweiler · 2011-02-18, 19:40

Quote:

Originally Posted by gd_barnes

Max is incorrect. I've always bought boxes without power supplies and then bought separate power supplies. I would not trust a pre-installed one. I've usually looked for middle-of-the road power supplies; not the cheapest; not the most expensive. But for these 2 most important boxes, high-end ones would be better.

Hmm, I see...I didn't know that. How much did you usually end up paying for the power supplies?

vaughan · 2011-02-19, 09:08

I'm buying the Professional Gold series power supplies from Corsair in recent months. They are Gold certified (high efficiency) and they are modular so I don't have my crunchers filled with unnecessary cables. The 750W version is powerful enough except for multi GPU machines.

gd_barnes · 2011-02-19, 09:11

Quote:

Originally Posted by mdettweiler

Hmm, I see...I didn't know that. How much did you usually end up paying for the power supplies?

$25-35. They are generally 470-500 watts. The cheapies are $18-25 at the store that I frequent.

You should know. When you originally sent me the specs for my first 6 AMD machines, you had power supplies with them.

em99010pepe · 2011-02-19, 09:46

You should be spending like $100 per power supply.

mdettweiler · 2011-02-19, 20:53

Quote:

Originally Posted by gd_barnes

$25-35. They are generally 470-500 watts. The cheapies are $18-25 at the store that I frequent.

You should know. When you originally sent me the specs for my first 6 AMD machines, you had power supplies with them.

Yes, but I knew that for the Intels you bought the power supplies locally; I didn't know what grade they were (some computer stores might not even stock the totally bargain-basement ones).

Yeah, $25-35 is still pretty cheap. That kind of price tier is generally OK for less-crucial stuff like crunching boxes, but for a server we'll want something better. Thanks Nugget and Vaughan for the recommendations--I'll do some shopping around to find a better replacement for jeepford (and also for Gary's i7 while I'm at it). Even if both of their existing power supplies are actually OK, it will be good to have better ones in there going forward and he can always use the two leftover cheapies as spares or for future crunching boxes.

gd_barnes · 2011-02-25, 22:26

The server machine will be down for about 30 minutes while I run some updates and perform some maintenance.

gd_barnes · 2011-02-25, 22:54

The server is back up now. Everything went smooth.

mdettweiler · 2011-04-26, 21:00

All noprimeleftbehind.net services seem to be down right now. No information thusfar on the cause of the problem.

Oddball · 2011-04-27, 05:21

Quote:

Originally Posted by mdettweiler

All noprimeleftbehind.net services seem to be down right now. No information thusfar on the cause of the problem.

Every outage is an outrage!

BTW, has there ever been any loss of data from these problems? It would suck to have a hard drive crash and lose all of the LLR residues.

AMDave · 2011-04-27, 06:45

Yes. Losing 25,862,597 residuals (to-date) would indeed suck.

Thus we have a DRP based on nightly off-server backups.
I have successfully tested the recovery from these backups.
I have even documented and tested a complete server s/w re-build from scratch.

RPO based on Maximum loss is 1 day plus 30 minutes (per scheduled backup propagation)

RTO is less than a week.

My target is 48 hours.

My tested recovery time was 5 hours from response time depending on hardware availability.
I made hardware available.
Hardware availability and config took longer than expected on the last test run, but the stats code has been upgraded to run on the newest versions of the middle-ware since that issue was identified so that problem has been eliminated.

Last DRP test run date was ... rummaging for the accurate date time ... 17-MAR-2011

Next DRP test run should be around September.

/edit - I checked the last off-site backup, it's getting a bit old. I'll archive a new one in the next 24 hours - edit/

AMDave · 2011-04-27, 07:06

BTW the outage appears to be over.
I logged in and checked the procs
Gary / Max has resumed the ports, both prpnet and llrnet
I also verified the scheduled stats tasks are running as scheduled.

"2011-04-27 02:03:48 and all is well, Oh Yae!"

/EDIT - that was server time, not post-time -EDIT/

2011-04-27, 06:45	#43
AMDave Jan 2006 deep in a while-loop 2×7×47 Posts	Yes. Losing 25,862,597 residuals (to-date) would indeed suck. Thus we have a DRP based on nightly off-server backups. I have successfully tested the recovery from these backups. I have even documented and tested a complete server s/w re-build from scratch. RPO based on Maximum loss is 1 day plus 30 minutes (per scheduled backup propagation) RTO is less than a week. My target is 48 hours. My tested recovery time was 5 hours from response time depending on hardware availability. I made hardware available. Hardware availability and config took longer than expected on the last test run, but the stats code has been upgraded to run on the newest versions of the middle-ware since that issue was identified so that problem has been eliminated. Last DRP test run date was ... rummaging for the accurate date time ... 17-MAR-2011 Next DRP test run should be around September. /edit - I checked the last off-site backup, it's getting a bit old. I'll archive a new one in the next 24 hours - edit/ Last fiddled with by AMDave on 2011-04-27 at 07:45

2011-04-27, 07:06	#44
AMDave Jan 2006 deep in a while-loop 2×7×47 Posts	BTW the outage appears to be over. I logged in and checked the procs Gary / Max has resumed the ports, both prpnet and llrnet I also verified the scheduled stats tasks are running as scheduled. "2011-04-27 02:03:48 and all is well, Oh Yae!" /EDIT - that was server time, not post-time -EDIT/ Last fiddled with by AMDave on 2011-04-27 at 07:21

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
What exactly is sent to the server?	paul0	NFS@Home	2	2015-03-12 23:00
Server bug -- new?	Christenson	Information & Answers	5	2011-07-12 21:44
Server Down?	Grant	Information & Answers	13	2008-11-24 19:37
New ECM-server available	andi314	Factoring	3	2003-08-31 11:22
New Server Hardware and price quotes, Funding the server	Angular	PrimeNet	32	2002-12-09 01:12

2011-02-19, 09:08	#35
vaughan Jan 2005 Sydney, Australia 5×67 Posts	I'm buying the Professional Gold series power supplies from Corsair in recent months. They are Gold certified (high efficiency) and they are modular so I don't have my crunchers filled with unnecessary cables. The 750W version is powerful enough except for multi GPU machines.

2011-02-19, 09:46	#37
em99010pepe Sep 2004 B0E₁₆ Posts	You should be spending like $100 per power supply.

2011-02-25, 22:26	#39
gd_barnes May 2007 Kansas; USA 3·7·491 Posts	The server machine will be down for about 30 minutes while I run some updates and perform some maintenance.

2011-02-25, 22:54	#40
gd_barnes May 2007 Kansas; USA 3×7×491 Posts	The server is back up now. Everything went smooth.

2011-04-26, 21:00	#41
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 14151₈ Posts	All noprimeleftbehind.net services seem to be down right now. No information thusfar on the cause of the problem.