Adaptec 51645 CLI and issues

Historically speaking – you could quite easily draw a parallel between the phrase “No one has ever been fired for buying IBM” and use of an Adaptec card for your storage array. In fact you could go as far as saying it would not be a SCSI card – but an Adaptec card.

They even bought out IDE controllers – while specifically ‘not IDE’ as their connectors were for one device and one device only – they were credible things. CPU, memory, battery, real… but they were rock solid.

Then came SATA – and they sadly fell off a cliff. The name became synonymous with array issues, failures and data loss. With 3ware disappearing into the scenery and LSI stepping up to the pitch. More over they have been bought and NOT maintained their name. This speaks of many loud bad things to me - and Micro Semi - it sounds like a geek getting over excited about a new shiny. Wrongness ^2.

SO. Here we are. Again. With our reputation...

This is a small post about extracting myself from a world of poo involving an Adaptec 51645 sporting a SAS backplane and 16 15k sas drives of live data in RAID 10.

I had available to me at the time two sets of tools, as the OS dependant IP management tool provided (storage manager) was unable to connect.

acli – for your rocket scientists;

ucli – more limited, but right for this job.

The moral of this story is a simple one – reported alarms on drive failures appear to be consistently lies. LIES you hear me!

Rescan device, finds drive healthy again, however fails it now. Removes drive. Adds drive. “Oh a nice new drive, allow me to prep this and re add it”. Nice one son.

This no bad thing – as maintaining a 600G 15k SAS array is now becoming an expensive past time with drives at £250+vat at the moment.

This example is from two Open-E patforms. At the terminal CTRL-ALT-R select the controller in question – and for your own sanity – select the ucli.

Turning off the alarm on Adaptec 51645

Probably a priority for you – as its really very loud.

For me and this instance it precompletes with the first part of the command – as that makes sense.

> acrcconf

For example. However for completeness here however I will give the complete command.

The help suggests - which is great if you know what they are – but there is no list. It enumerates from 1 it turns out – so one controller is number one – happy days.

> arcconf setalarm 1 silence

...and relax. We can all think a little better now.

Show the configuration on Adaptec 51645

Your best go to is not so much the logging, but the configuration.

This was initially a little counter-intuitive.

> arcconf getconfig 1

Shows the configuration that is currently in play. This shows logical devices, drive groups, physical devices, backplanes, stuff. It should be quite obvious from here as to what has gone wrong.

Chances are – it will have what you need – unless there is a drive showing as dead and dead.

You can narrow this down to specific parts as follows:

> arcconf getconfig 1 pd

> arcconf getconfig 1 ad

...and so on. But not specifying anything is dandy – as long as you can scroll back through it all.

Show logging and events on Adaptec 51645

This is not as helpful as you would probably think.

It certainly does not feel the need to conform to any standards with a date and time on the left.

You can however extract the important things of what, when, and where from the logging.

The two I have found the most useful up to this point are:

> arcconf getlogs 1 device

Which will return device issues that have been reported – for example – here we have an obfuscated returned example. This returns data relating to the specific device.

driveErrorEntry smartError="false" vendorID="SEAGATE " serialNumber="3SL99DHN123456789ABC" wwn="9876c500241d1234" deviceID="53" productID="ST360005" numParityErrors="0" linkFailures="0" hwErrors="1" abortedCmds="0" mediumErrors="8" smartWarning="0"

Lets use something a little more specific – a little more why am I here, lets make the bad stop...

> arcconf getlogs 1 dead

deadDriveEntry vendorID="SEAGATE " rtcSeconds="30" rtcMinutes="42" rtcHours="4" rtcDay="6" rtcMonth="12" rtcYear="2016" serialNumber="3SL99DHN123456789ABC" productID="ST360005" wwn="9876c500241d1234" failureReasonCode="4"

The time is in there if you look. Honest. So you can see when this event occurred. Hopefully if this is a redundant platform – this will be something you have to look up as opposed to the specific point that all the alarms in your world started going off.


There is no cake.

Chances are there is no drive issue either.

Identify the drive in question (you can use the blink option – however YMMV depending on the backplane in use – for me the lights went out!) - and keep an eye on it’s behaviour. A red light is pretty obvious – however some time this is not the case.

The controller simply has not had an answer back from the drive fast enough for it’s liking – and as a result it is end of days, and Goodnight Vienna for that drive.

> arcconf rescan 1

Let it check over the drives it has available – and then run the config command again to see if it sees the drive.

> arcconf config 1

The kinds of things you are looking for are both in physical drives, groups (which I found the most obvious) or Controller Information at the top of that output – things like this:

 Defunct disk drive count                 : 0
 Logical devices/Failed/Degraded          : 1/0/1

No really – its all good!

There is a fair chance that this is just choosing to ignore you. So we wrestle the drive again.

The channel and id are shown in the physical ID’s in the format of (0,53) for example – given the

> arcconf tast start 1 device CLEAR

> arcconf task start 1 device VERIFY

> arcconf rescan 1

> arcconf getconfig 1

...and compare notes.

The hateful Java based GUI

There is the hateful Storage Manager to drop back to – yay.

I am giving the above as for this learning curve this would not connect.

Looking at that page – the majority of those links are for drivers for various OS’ that will already be aware of the score… what you are looking for there is the Storage Manager link – giving you a choice of Windows or RedHat/CentOS/Fedora downloads.

Great news if you are running one of those, or are running a VM with one of those Happy Days. If you are running a Debian/Ubuntu or something .deb based – then installing ALIEN will allow you to convert that from a .RPM to .DEB and install this. Using the –scripts switch however for alien resulted in badness – although might install it better. None the less, called by command line – progress. Or, in this case, reinforcing the point we cannot connect using this method – hence CLI learn and share.

One Response to “Adaptec 51645 CLI and issues

  • anthony
    7 years ago

    This particular adventure took place, and has taken place since on an Open-E platform.

    From their console there is a login that required shortcut keys to get anywhere.

    Open-E key combination for RAID controller set up is:

    CRTL + SHIFT + R

    PuTTY apparently does nice pass through according to their support – but I have had no issue with standard SSH console.

    Setting up the console can be found under:


    For me this saves a world of pain as these are old school Supermicro chassis – with KVM on the back only. Arse. Equally listening to the alarm is a dream behind the machine.

Leave a Reply

Your email address will not be published. Required fields are marked *