Adaptec have released a set of new "Low cost" hardware raid cards. The 2405 and 2045. Which they say is based on the 5805 etc. See details here
While the latest (April) VMware HCL NOT stating the 2x05 cards (2445 is there but this is a nonexistent card and it's not the 52445 which is already listed and a very different beast.....) while the Adaptec site specifically states that ESX 3.5 is supported.
Given that in other posts I have seen issues about the 3405 cards (and experienced an issue with the 3805 myself) I'm wary of trusting just Adaptec alone. We are looking to use these cards in small branch office servers that have a low requirement for storage (eg 2 146GB SAS will be fine) or as low cost entry for clients wanting to try VMware Foundation before committing to the sometimes significant hardware costs/upgrades.
So my questions are:
Is anyone successfully using the 2x05 cards with either SAS or Sata, and have they had any issues with them?
Does anyone at VMware care if a manufacturer states their product is compatible with ESX, without VMware's approval (assuming the 2445 entry in the HCL is not just a typographical error for 2445/2405)?
I haven't used the cards, but it's a bit odd that Adaptec's site lists them as supported for ESX 3.0.x when support for dual mode SAS/SATA controllers was only started with 3.5. It may be that the card is close enough to a SCSI cousin so that ESX thinks that it is a SCSI controller.
Given that you're only looking for RAID1 have you considered just getting a supported server with an onboard RAID controller?
Yes I had considered this, but I like the "Unified" nature of the Adaptec line. It means we can shift disks between controllers and retain the Raid information(500GB takes a long time to transfer even at 200Mb/s), and use the same raid tools, snmp monitoring etc. I have my staff to think of as well, I'd rather they knew a range of cards back to front, than try and deal with the intricacies of Onboard variations. It also means (and again I know this is not supported by VMware) we can occasionally reuse the hardware they have and prove the consolidation aspect. eg rather than spend $4000 on a new server, they spent the same amount on a new controller, new disks, VMware and MS and our consulting. This gets our foot in the door and gets them thinking about the possibilities (this is the SME market of course
But, getting back towards topic, it is interesting that the Series 5 "Range" have been certfied by AppLabs(according to Adaptec), yet if you go there and search for "esx", no hits, but you can't hide anything from google! It finds their(Applabs) press release here so I expect that is how they(Adaptec) justify the 3.01 claim and then because they state that the 2405 cards are based on the same chipset as the Series 5, they can infer the same. Also it does not say directly that SATA is certified, but it does say "VMware Server ESX 3.02 and 3.5, such as expanded storage options and SATA support"... pretty ambiguous.... Since Applabs is listed by VMware as a certified testing centre... it must be right.....although I would have thought one of the VMware requirements would be to report the results of testing, but I guess listing in the HCL is enough......
Adaptec is using a unified device driver for VMWare. The aacraid.sys which is included in VMWare does support all families of Unified Serial controllers. This includes Series-3, Series-5 as well as the new Series-2. The Series-2 might not be listed on the official HCL as of yet but as it uses the same driver which is in VMWare for long it should work fine.
BTW: Adaptec has the same unified driver concept for Linux.
Thread moved to the VI: ESX 3.5 forum
VMware Communities User Moderator
Well in theory, what you say is correct, and also why I have stuck with the Adaptec cards. The "Unified" nature is very attractive, and being able to shift Raid Arrays from one server to another. But having said that, I've been burnt way too many times by assuming that just becuase a manufacturer implies one card is the same as another, then they are.
A classic example comes to mind of another storage company, who released a Sata "Plus" card, so the Model name was "Sata Blah Blah Plus". One would assume that it did the same things as the previous card... "Plus" a bit extra.... nope. What it did was add an extra port (this was the "Plus" bit) and then remove all of the Raid1 and Raid1+ functionality that made the previous card work really well.
Now I know that this was not Adaptec, and it was a low cost card (sub $100) but if I thought more about it I am sure I could find more to illustrate my point.
I decided to bite the bullet and orderd a 2405,(since we can't get stock of anything else at the moment anyway...) we installed it in place of a 3805 we wre having major dramas with, and touch wood... no PSOD's it just worked straight out of the box.... Raid1, 2x 300GB Segate SAS and is going well so far.....
So... unofficially it looks like it does work.(Using SAS drives at least.)
>Is anyone successfully using the 2x05 cards with either SAS or Sata, and have they had any issues with them?
yes, i got a 2405 today and trying esx 3.5 on that - installs fine, vm`s run fine - and after an hour i got a posd (purple screen of death) with a message like this:
i`m updating to latest patches now and then will check for the controller/COS irq thingie.....
Damn! now I am nervous! The 2405 has run like a champ for us so far. so the error you got was "Heartbeat" related?
The only other error we have received, and can duplicate, is during shutdown we get a PSOD as well, but it does not mention "Heartbeat".....
What motherboard are you using?
>Damn! now I am nervous! The 2405 has run like a champ for us so far. so the error you got was "Heartbeat" related?
that`s the issue with unsupported/not-certified setups
yet it was like this (got that two times):
[31;1m0:02:53:05.285 cpu1:1025)ALERT: Heartbeat: 470: PCPU 0 didn't have a heartbeat for 182 seconds. may be locked up [0m
[31;1m0:02:53:05.285 cpu0:1064)ALERT: NMI: 1625: Faulting eip:esp 0x63fa9b:0x3aa3e64 [0m
0:02:53:05.285 cpu0:1064)0x3aa3e88:0x63fa9bUtil_Udelay+0x5a stack: 0x5, 0x89d6a5, 0x4020
0:02:53:05.285 cpu0:1064)0x3e241088:0x89d6b4aacraid_esx30+0x76b3 stack: 0x0, 0x0, 0x0
0:02:53:10.213 cpu1:1061)LinSCSI: 3201: Abort failed for cmd with serial=2145, status=bad0001, retval=bad0001
ah - just see this one in the log from the other post:
0:02:53:11.086 cpu0:1064)<3>aacraid: aac_fib_send: first asynchronous command timed out.
Usually a result of a PCI interrupt routing problem;
update mother board BIOS or consider utilizing one of
the SAFE mode kernel options (acpi, apic etc)
that message didn`t exist in my PSOD.
>What motherboard are you using?
it`s a lenovo m57 - don`t know what type of mainboard is inside. you may check the specs or i can take a look next week.
overall, the server with 2405 card is running fine so far (just test vms) - but i/o freezes from time to time (about some weeks of runtime).
i`m curious if this is a problem with the vmware driver or with the adapter firmware - or both.
anybody using this controller?
i think i contact adaptec for this....