HP p410i w/ 512mb cache and SATA drive

[bmatic]

Hello. First post here. I looked through the forum but couldn't find exactly what I was looking for.
I recently purchased an HP DL380g6 server which came with the HP P410i RAID controller with 512mb of cache and battery backup. I currently have 4 x 146gb SAS drives in there with the intention of adding another 4 drives, which leads me to my question.
I could get some 300gb SAS drives, but for a few more dollars, I can also get HGST Travelstar 7K1000 2.5-Inch 1TB 7200 RPM. I read that this particular HP controller doesn't play well with non-HP drives. Has anyone had experience with this raid controller and non-HP drives? I don't want to get these drives and then have my array mess up.

Thanks for everyone's input.

 

[5teve]

I'm running HP P420 and HP822 all with hitachi ultrastars and some crappy old samsungs with no issue (relating to drives anyway) I'm even running the cards in non hp machines. I'm fairly sure the ML350G6 I have access to has a P410 in and is using non HP drives also.

You may want to check your firmware level also as HP are apparently opening up their devices due to a licencing agreement with PMC(?) so later firmwares may help.. I'm sure Patriot will chime in shortly tho to give you a firm answer

Steve

 

[vanfawx]

Yup, we've used crucial m500's with no problem on that controller! This is on a HP DL360 G6.

t.

 

[bmatic]

I'm running HP P420 and HP822 all with hitachi ultrastars and some crappy old samsungs with no issue (relating to drives anyway) I'm even running the cards in non hp machines. I'm fairly sure the ML350G6 I have access to has a P410 in and is using non HP drives also.

You may want to check your firmware level also as HP are apparently opening up their devices due to a licencing agreement with PMC(?) so later firmwares may help.. I'm sure Patriot will chime in shortly tho to give you a firm answer

Steve

Thanks for your response... speaking of upgrading the firmware, how would I do that? The HP site with all the different SAS/SATA/Controller FW is confusing to me. I have vSphere 6.0 installed on the server and running a few VM's on it. What is the best way to upgrade the p410i firmware? Which file would I need to download? Can it be updated using USB flash drive?

Thanks!

 

[bmatic]

Yup, we've used crucial m500's with no problem on that controller! This is on a HP DL360 G6.

t.

Thanks for your reply.

 

[5teve]

Thanks for your response... speaking of upgrading the firmware, how would I do that? The HP site with all the different SAS/SATA/Controller FW is confusing to me. I have vSphere 6.0 installed on the server and running a few VM's on it. What is the best way to upgrade the p410i firmware? Which file would I need to download? Can it be updated using USB flash drive?

Thanks!

Firmware upgrade on the P420 and P822 was just downloading the executable (for windows - sorry) and starting it.. it extracts it, flashes it and is complete.. very painless.

From what I can see.. wait for someone else to confirm as I am very very very new to this and only know windows and limited freenas (I'm a tinkerer not a professional) .. you need the following Drivers & Software for HP P410 with 512MB Flash Backed Cache Controller - HP Support Center.

Expand the storage controller option - and there is only one file available - version 3.3.. you may want to confirm what you are running currently first tho.

Alternatively.. if you have some bays free.. why don't you just borrow / beg a non hp drive and try it?

Steve

 

[50chickens]

I have a HP DL360 G6 with an onboard p410i controller and have some problems with the performance of exactly the drives you're thinking about using - HGST Travelstar 7K1000. The server runs fine using the 300GB SAS drives but using SATA drives (the below benchmarks apply to the WD 750GB scorpio black (WD750BPKT) i get low transfer rates compared to the p410 add-in card.

this is the transfer rate using iometer when attached to the p410 add-in card (3 x 1tb hitachi drives in raid 0, 256k stripe, 75%/25% write/read cache ratio, Drive write cache enabled). This is attached as a non boot drive so that there's zero io going to it except the testing. the 350mb/sec is around what you'd expect for this configuration. the write test is 1mb block size, 100% sequentual writes test using iometer (1 worker).



this is the same 3 drives attached to the onboard p410i - same raid configuration, internal bays used and benchmark configuration. As you can see the average throughput is low compared to the onboard and it's very bursty.



So as far as i can tell there's some difference between the two. the p410i has throughput problems with sata drives and the p410 add-in card does not. I have upgraded both to the latest firmware (6.62 at the time of writing), both have the same amount of ram (512mb) and checked all the options like drive write cache, cache ratio, stripe size etc are all the same.

I also have this same io problem for a DL 380 G6 using the same type of sata drives plugged into the onboard p410i controller (although it's a slightly older firmware version).

 

[mrkrad]

power management disabled? (MAX,MAX,MAX performance in bios?)

 

[Robert Franz]

I have a HP DL360 G6 with an onboard p410i controller and have some problems with the performance of exactly the drives you're thinking about using - HGST Travelstar 7K1000. The server runs fine using the 300GB SAS drives but using SATA drives (the below benchmarks apply to the WD 750GB scorpio black (WD750BPKT) i get low transfer rates compared to the p410 add-in card.

Was the array built before or after the fw flash?

I'm doing some testing/rebuilding of the storage in a g7 with a p410i 1g.

It was originally running raid 6 using a trial key.

Sent to the colo and the trial key forgotten about until it returned to the mothership.

I read - in this thread, I think, about the additional raid levels being unlocked with a fw flash, so it seemed like a no brainer to go ahead and flash it.

I did, then removed the sas drives and put in 5 Samsung Pro 1 G drives in raid 5.

Format started uneventfully, I exited the ACU and let it continue to build.

This is a vsphere host - 5.0 at that time.

I went and did some other things, came back ran a quick benchmark and saw result around the throughput limit for that card when using parity.

I wanted to double check the cache setting, and probably tweak it a bit based on my expected work loads, so back to ACU.

Everything looked fine, until I drilled down into the array and it indicated that parity had failed to build.

I'm not a storage guy, so I had to think about that for a moment.

Yeah - it tracked with what I was seeing - raid 5 before parity is built is more or less raid 0, and since parity was never built in the first place, the array would not go into degraded mode since there was nothing to rebuild - sort of.

At this point, I had already flashed to the current firmware, but I was also not seeing the raid 6 option.

I wasn't planning on running raid 6 anyway due to it's poor write performance, but it was sort of a miner's canary that told me something was up.

I deleted the trial key that had expired 3 years previously and rebooted - no joy.

Shrugged and moved back off the 1 vm I had migrated to it, deleted the array and rebuilt it.

Oh, and just to confuse the issue - possibly because it was 2am - I also updated from esxi 5.0 generic to 5.5 HP oem to make sure I hadn't missed and drivers or created a driver situation that was jacked up but that I wouldn't see.

NOW it showed that it was built properly.

Also, I now started to see the storage controller in esxi in the gui, where previously I had not.

It may have been exposed at the command line level - but I wasn't the point of looking there since a rebuild is quick and my esxi cli skills are minimal.

This is kind of a long way around to suggesting you try a simple rebuild if the array was initially configured on previous firmware, and also to double check the drivers in use - low level firmware drivers do take input from OS drivers, so an old/incorrect/broken driver can impact the way an array is built or is accessed.

And check the array in ACU.

I did also look at it via various other utilities and none of them indicated an error.

For that matter, it was odd that ACU wasn't screaming on the main page - failure to initialize parity is a pretty serious condition - so do make sure you drill down into the array and verify it doesn't show something off there.

I did see specific errors in the Insight diagnostics that run from the 8.whatever image that's used for new server setup, indicating hot removal of multiple drives.

I did move a drive initially after power up, but well before post and this was before I had booted to ACU to create the array initially - also I know I only moved one, so I think that's unrelated.