Software RAID1 corruption. [Necro question, maybe]
Hi, Thanks for getting to my post in your busy day (and night).
I came across the board researching a problem I have with my file server. http://icrontic.com/forum/showthread.php?t=2534
My problem is a little different so I did not want to hijack the thread.
I have an NFORCE2-based Shuttle AN35 Ultra board. I only got this board last year. Before then, I was using a Fujitsu-Siemens Scenic L P3-1Ghz. Pretty slow, eh?
Anyway, I had a software RAID1 with Windows. The PCI controller is a Silicon Image 3112 with the 4.2.84 firmware (latest) and latest driver from Microsoft Update (MS is x.71 and SI is x.68).
Periodically the file server had win32k.sys crashes, which bugged the crap outta me, until I removed a remote control software package and these went away. Naturally, every time my raid began rebuilding itself.
I was also using 8rdavcore to dynamically change the FSB from 166 (rated for the Athlon XP 2600+) to 100, with the 166(333) timings of 7-4-4-2.5. The memory is a pseudo-paired (same rating, same manufacturer) TEAM Value RAM. The Shuttle has never had a problem with this RAM. 8rdavcore only changed the FSB and CPU voltage, 166 to 100 and 1.675 - 1.30, and back again during high demand (above 83%). The processor is an Athlon XP production date 0343, so unfortunately it is clock-locked.
Very recently, some kind of corruption of both the second and third paritions of the RAID1 happened. The machine rebooted, and drives D and E are now inaccessible. The raid1 is on Seagate 750GB drives, both connected directly to a SI3112 PCI card.
I thought it could be related to a bus speed divider error. Now I read the post above, and it seems this could not be the case. Whew!
I tried chkdsk D: /R, also E: btw, and this does not recover the partition. If you have an idea what I could do, please let me know also. I can change drive letters to it, and that is about all. C: seems fine. C D and E are on the same RAID1.
After reading a few days ago, shortly after the corruption, I thought having set the bus speed at 150 created an overclocked PCI bus, causing the disk corruption. Now, after reading assertions here, I do not know where to look. The Si3112 has a 'hardware' raid option, although the actual computation of the RAID1 information is done on the host CPU, so for flexibility the system was using software (Windows). Maybe there is a better way to setup RAID1? I have seen an external box for about 114 for two SATA drives, which has built-in RAID1. It is a FANTEC MR-35DUS2.
I currently have two hard drive housings, eSATA MAPOWER cases, unused, because the PCI card Si3112 has an internal SATA connector on the outside (on the bracket, where an eSATA connector should go), in addition to the 2x internal SATA connectors. Not knowing much, I thought this connector was a normal eSATA, until I tried to plug one of the MAPOWER cases in.
Maybe a different RAID adapter is called for, or an external raid solution? What do you think caused the glitch? Is there a solution to recover the data (simple)?
It's not all terrible news about the data loss: I happened to recieve an RMA'd drive back, one of the 750GB Seagates. A 500GB Seagate temp replacement was swapped out a week ago, and the replaced 750GB Seagate swapped in, in the existing RAID1. The 500GB is just sitting, patiently now, in a cabinet. The partitions on the 500gb are undamaged, and I can rebuild a RAID1 from the information. I also have a backup of the most essential data on an external 250GB PATA/USB Seagate from the same time.
In retrospect I think I need a better RAID1 card. I can't spend 200 Euros for one, we just do not have it in the budget. Prices for what I could get are here: http://www.csv.de [Klick on Komponents, Festplatte, Controller]. Because the Shuttle AN35 does not have a facility to regulate the fan speeds, and it's kinda a bare-bones board anyway, I recently bought (a few minutes ago) an MSI K7N2 Delta2 Platinum on ebay. This has better fan regulation and generally is a better board (I think, I don't know, what do you think?) I also picked up yesterday on ebay a 0310 production date Athlon XP 2500+, so I can mess with changing the dividers. Is this divider changing possible on these boards?
I came across the board researching a problem I have with my file server. http://icrontic.com/forum/showthread.php?t=2534
My problem is a little different so I did not want to hijack the thread.
I have an NFORCE2-based Shuttle AN35 Ultra board. I only got this board last year. Before then, I was using a Fujitsu-Siemens Scenic L P3-1Ghz. Pretty slow, eh?
Anyway, I had a software RAID1 with Windows. The PCI controller is a Silicon Image 3112 with the 4.2.84 firmware (latest) and latest driver from Microsoft Update (MS is x.71 and SI is x.68).
Periodically the file server had win32k.sys crashes, which bugged the crap outta me, until I removed a remote control software package and these went away. Naturally, every time my raid began rebuilding itself.
I was also using 8rdavcore to dynamically change the FSB from 166 (rated for the Athlon XP 2600+) to 100, with the 166(333) timings of 7-4-4-2.5. The memory is a pseudo-paired (same rating, same manufacturer) TEAM Value RAM. The Shuttle has never had a problem with this RAM. 8rdavcore only changed the FSB and CPU voltage, 166 to 100 and 1.675 - 1.30, and back again during high demand (above 83%). The processor is an Athlon XP production date 0343, so unfortunately it is clock-locked.
Very recently, some kind of corruption of both the second and third paritions of the RAID1 happened. The machine rebooted, and drives D and E are now inaccessible. The raid1 is on Seagate 750GB drives, both connected directly to a SI3112 PCI card.
I thought it could be related to a bus speed divider error. Now I read the post above, and it seems this could not be the case. Whew!
I tried chkdsk D: /R, also E: btw, and this does not recover the partition. If you have an idea what I could do, please let me know also. I can change drive letters to it, and that is about all. C: seems fine. C D and E are on the same RAID1.
After reading a few days ago, shortly after the corruption, I thought having set the bus speed at 150 created an overclocked PCI bus, causing the disk corruption. Now, after reading assertions here, I do not know where to look. The Si3112 has a 'hardware' raid option, although the actual computation of the RAID1 information is done on the host CPU, so for flexibility the system was using software (Windows). Maybe there is a better way to setup RAID1? I have seen an external box for about 114 for two SATA drives, which has built-in RAID1. It is a FANTEC MR-35DUS2.
I currently have two hard drive housings, eSATA MAPOWER cases, unused, because the PCI card Si3112 has an internal SATA connector on the outside (on the bracket, where an eSATA connector should go), in addition to the 2x internal SATA connectors. Not knowing much, I thought this connector was a normal eSATA, until I tried to plug one of the MAPOWER cases in.
Maybe a different RAID adapter is called for, or an external raid solution? What do you think caused the glitch? Is there a solution to recover the data (simple)?
It's not all terrible news about the data loss: I happened to recieve an RMA'd drive back, one of the 750GB Seagates. A 500GB Seagate temp replacement was swapped out a week ago, and the replaced 750GB Seagate swapped in, in the existing RAID1. The 500GB is just sitting, patiently now, in a cabinet. The partitions on the 500gb are undamaged, and I can rebuild a RAID1 from the information. I also have a backup of the most essential data on an external 250GB PATA/USB Seagate from the same time.
In retrospect I think I need a better RAID1 card. I can't spend 200 Euros for one, we just do not have it in the budget. Prices for what I could get are here: http://www.csv.de [Klick on Komponents, Festplatte, Controller]. Because the Shuttle AN35 does not have a facility to regulate the fan speeds, and it's kinda a bare-bones board anyway, I recently bought (a few minutes ago) an MSI K7N2 Delta2 Platinum on ebay. This has better fan regulation and generally is a better board (I think, I don't know, what do you think?) I also picked up yesterday on ebay a 0310 production date Athlon XP 2500+, so I can mess with changing the dividers. Is this divider changing possible on these boards?
:thumbupThanks for any, or all of, your responses!!!:thumbup
0