CaptainSpalding Posted March 18, 2010 Author Share Posted March 18, 2010 Now THAT points to bad memory! Do you mean RAM? If you refer to the HDD memory, this is second brand new HDD with bad memory.... :'( Link to comment
CaptainSpalding Posted March 18, 2010 Author Share Posted March 18, 2010 RAM Thanks, will get new ones tomorrow. Link to comment
purko Posted March 18, 2010 Share Posted March 18, 2010 RAM Thanks, will get new ones tomorrow. No need to wait till tomorrow to prove/disprove my theory. Reboot your server, and from the boot menu on the console select Memtest86+ You should be able to run it for at least a few hours before you believe you have good RAM. How many sticks of RAM do you have in your server? If more than one, then start unplugging them one by one untill Memtest86+ can run stable. This way you can isolate a single RAM stick that's causing you the problem. Link to comment
CaptainSpalding Posted March 18, 2010 Author Share Posted March 18, 2010 I've got two 1Gb sticks and running the memtest on 1 of those now. Thanks! Link to comment
purko Posted March 18, 2010 Share Posted March 18, 2010 I've got two 1Gb sticks and running the memtest on 1 of those now. Did memtest bomb when both of them were in your server? Link to comment
CaptainSpalding Posted March 18, 2010 Author Share Posted March 18, 2010 Just ran 5 passes with no errors with other of the disks... Might not be the ram? Link to comment
purko Posted March 18, 2010 Share Posted March 18, 2010 Just ran 5 passes with no errors with other of the disks... Might not be the ram? Passing memtest with just one stick at a time is not enough. You should be able to pass memtest for a few hours with you final RAM configuration. You mentioned that you would use 2 sticks, right? Well then, put them both in, and see if it'll pass. Some RAM sticks may be OK by themselves, and yet cause trouble when used together. Link to comment
CaptainSpalding Posted March 19, 2010 Author Share Posted March 19, 2010 I was little ahead of time and last night I ran 9 hours of memtest with both sticks and 0 errors. EDIT: Thanks you for your help. I now changed the sata cable and it works. Damn that was a quality cable and not DOA... could not expect. Link to comment
CaptainSpalding Posted March 19, 2010 Author Share Posted March 19, 2010 Hmmm... out of couriosity I checked the latest syslog and it says: Mar 19 11:30:14 Tower kernel: Phoenix BIOS detected: BIOS may corrupt low RAM, working around it. What does that mean? syslog.txt Link to comment
Kaygee Posted March 19, 2010 Share Posted March 19, 2010 There is a known issue with this Phoenix bios, someone has patched it and this message lets you know you wont have this issue. Link to comment
CaptainSpalding Posted March 19, 2010 Author Share Posted March 19, 2010 There is a known issue with this Phoenix bios, someone has patched it and this message lets you know you wont have this issue. Great, thanks! Link to comment
CaptainSpalding Posted March 19, 2010 Author Share Posted March 19, 2010 Can data-rebuild really take 1500 mins? I have 12 drives including parity. Q6600 and 2Gb of RAM. All of the HDD's are 1.5Tb. BTW, in the syslog was mentioned something about 1.5Gb, but all my drives are SATA II 3.0Gb? Link to comment
Joe L. Posted March 19, 2010 Share Posted March 19, 2010 Can data-rebuild really take 1500 mins? I have 12 drives including parity. Q6600 and 2Gb of RAM. All of the HDD's are 1.5Tb. BTW, in the syslog was mentioned something about 1.5Gb, but all my drives are SATA II 3.0Gb? Your disk controller negotiates the speed with the disks. The disks may be jumpered to operate at the slower more, or the disks controller may have found the higher speed not usable and dropped to the slower speed. That really does not matter since SATA-1 is still far faster than the sustained read or write speed to a drive. Most people report read top speeds between 75 and 100 MB/s on a drive, slowing to 60 MB/s or less on inner cylinders. Let's assume best case and it will be 1GB per 10 seconds, at 75MB/s it would be 13.3 seconds per 1Gig. Let's assume your motherboard has absolutely NO bottlenecks on its internal busses. Now, writing to a disk is typically slower. But for this exercise let's assume it can also be written to at 1GB per 10 seconds. (100 MB/s) You have 1500GB (1.5TB) to read, therefore, 15000 seconds at 100MB/s... Or 19950 seconds at 75 MB/s. Dividing by 60 gives between 250 minutes and 332.5 minutes as the absolute best case with no time to actually calculate parity to write. Remember, it will be slower than the time it takes to perform a full parity check. The rebuild will be limited by the speed of the slowest drive involved and/or the capacity of the system bus to read all the data. Don't forget you are reading 1.5TB each from 11 of your disks. 1500 minutes would indicate 1 minute per gigabyte, or 16.6 MB/s. Joe L. Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.