RAID1 to RAID5 Migration chugging along!

[mdmint]

Picked up a 3rd SATA 250GB today for my main workstation which is also being used for Library space. Was config'd RAID1, way cool on the fly didn't even power down just slide in the new HD pop into the util and fired off migration from RAID1 to RAID5! (Promise FastTak S150 SX4 Ctrl w/256M ECC paired with Promise Superswap 1100 cages left over from upgrading SageServer to Escalade 9500S-8 and Escalade cages) That'll up my library space from 200GB to 450GB. Later (soon) I'll add a 4th HD & expand the RAID5 for 700GB Library to augment my main 1.495TB Sage storage. That oughta hold me for awhile.

FWIW, it's 31% done migrating 3&1/2hrs later... Of course, it might be a bit further along if I was using the workstation! (comskip's been running on the workstation too)

[mdmint]

I discovered an issue with the JD model drives but not during Sage usage, works flawlessly for Sage duty even with 5 DVD standard records and playback simultaneously. My RAID5 on Escalade 9500S-8 using WD2500JD drives would get WinXP delayed write fatal error when backing up my workstation to the array. Research said turn of write back cache. Tried that and worked better but still not working. Western D makes SATA models specifically for RAID usage with a different limited error correction recovery timing, model WDxxxxSD. I pretty much decided I was going to go through the painful process of buying all new drives and selling the JDs on eBay. Called WD tech support inquiring about JD update for better RAID compatibility, got transferred to 2nd level support, no updates for JD models. However, was told WD will RMA replace the JD model drives with SDs at no uplift charge! Other than my cost to ship them back of course. Way cool on Western Ds part. First 5 advancement RMAs on the way. 5 to go. (Limit 5 advance RMA requests per day).

Interesting to note that while doing large file transfers to Escalade 9500S-8 RAID5 with WD2500JD array delayed write errors occur, doing large file transfers to Promise S150 SX4 RAID5 WD2500JD array delayed write errors do not occur. Seems the JDs are more compatible with Promise controllers than Escalade. (xfers over copper gigabit so HD subsytems not slowed by network) I'm RMA'ing the JDs in the Promise array anyway to be safe since I can.

[mdmint]

Yesterday my first 5 WD2500SD arrived. (free RMA upgrade to replace my 10 WD2500JD) First replaced on in my workstation running Promise S150 SX4 3 HD RAID5. Few minutes later replaced in Sage Server running 3ware Escalade 9500S-8 HD RAID5 w/online spare. (disabled online spare before replacing so wouldn't automatically start rebuild to that drive) Since both arrays using exact same original 250G JD drives and being replaced with exact same SD drives figured rebuild time ought to be about the same. Not quite apples to apples since Promise array 3HD and 3ware 7HD but since it take longer to write then read figured should be close. Not.

9500S-8 took 2hr 27min to rebuild 1st replaced HD.
when done replaced a another before going to bed while Promise still rebuilding 1st drive. Checked this morning and 2nd 9500S-8 replaced HD took 2hr 21min. Just replaced #3.

The S150 SX4 took 3hr 22min to rebuild.

FWIW, that's with 9500S controller background rebuild priority set 80% towards faster I/O vs Rebuild and Promise initally at 50% then upped it to 75% faster rebuild vs I/O after about 30% done. Would have thought the Promise set to higher rebuild priority would have been faster rebuilding than the 3ware but such was not the case. Of course, again not the same size RAID5 array and don't know how much difference that would make.

Bottom line is they both did/are doing the rebuild jobs fine without shutting down and with data available at all times. Sweet, gotta love hotswap bays!

[Watter]

Where did you get the case or enclosures for the hot swap drives?

[mdmint]

Originally Posted by Watter

Where did you get the case or enclosures for the hot swap drives?

NewEgg. Ideally you'd want to match the enclosure to the RAID controller for monitoring and notification etc. (I say ideally because when I bought mine in April 04 3ware advertised cage monitoring support coming in the Fall. But the got bought by AMCC and it hasn't happened and per tech support they don't know when or if now...)

With the 9500S-8 in Sage Server using two RDC 400 SATA 4in3 cages. http://www.newegg.com/app/ViewProduc...121-302&depa=0

With the S150 SX4 in my workstation I'm using 4 SuperSwap 1100s. http://www.newegg.com/app/ViewProduc...102-024&depa=0
At the time they didn't make the SuperSwap 4100 (4in3 cage) or I would have it gotten it instead http://www.newegg.com/app/ViewProduc...102-034&depa=0

Adaptec also makes cages to pair with their SATA RAID controllers.

[mdmint]

The first three drives I replaced on the Server I did when I knew nothing Sage wise would be going on. Just for fun, I fired off comskip from my workstation Client. Then went into 3ware interface and disabled the next drive to be replaced on the 9500s, replaced it and initiated rebuild. All the while comskip continued processing from the Client happy as a clam. Of course, the rebuild will likely take longer with constant file access vs idle. Will be interesting to compare.

[mdmint]

The last couple days I've been replacing my 8 WD2500JD drives in my 3ware 9500S RAID5 array with WD2500SD. The first 6 I did when I knew no Sage recording or viewing would be happening. Did do one with my workstation comskip processing for the heck of it no problem. This morning replaced the last HD in the array when I got up, the 7th (8th is online spare so doesn't affect performance.) While eating breakfast we watched a recorded episode of 30min Meals, SageServer array rebuilding (Std DVD Q). No playback problem. Cool. I was hoping it'd still be rebuilding when a 11:30am non-critical show (cooking show) was scheduled to record but it finished rebuilding too soon.