Areca 1880

[Nizzen]

Now I am testing Samsung f4 2tb in raid-5. So far so good Looks like we have a cheap winner for Areca

Not much performanceloss wit raid-5 vs raid-0. (read)
Firs a small array:

[Tiltevros]

Nizzen what is ur maximum Mb/s with 1880 and SSD in HD tach??

[Wintel64]

NEW beta firmware build WOOOHOO!!

ftp://ftp.areca.com.tw/RaidCards/BIO...a/Build101006/

After 3 weeks, does the beta firmware is stable?

[Computurd]

oh yea, no issues. seems the battery issue is fixed too.

[Wintel64]

oh yea, no issues. seems the battery issue is fixed too.

What is the battery issues?

[Computurd]

a rare occurrence thankfully. basically if for some reason you get a 'bad' initialization on your card, when the bios is booting up (which is also rare), you cant get the card to reboot until you remove the BBU, and then boot. it is like if there is a bad load of the card bios the BBU keeps the bad info or whatever in there even after you reboot. that is my hypothesis at least of what is going on with it. i have only seen one other user have it happen to him, so may be a certain 'like' combination of hardware.

[odditory]

you're right CT, i've had the battery issue too, once on each of my two 1880i's, to the point I just removed my BBU's.

i'm able reproduce it by the rare occasion that I have one of my Hitachi 1TB or 2TB drives show a "timeout" in the event log, which on rare occasions occurs when an array is waking up from being spun down. the whole areca card locks up and the server has to be hard reset. then unless I remove the battery the areca won't init its bios.

the issue appears to be a combination of setting too low of a spinup time (i.e. 0.4 or 0.7 seconds) in HDD power management in the Areca webgui, and certain models of harddisks. the problem used to be WAY worse on the 1680 series cards, and has all but gone away on the 1880's. its also not limited to the Hitachi drives (i've seen it happen on some seagates too). i'm still trying to narrow down exactly what causes it, areca's been no help and says "the drive isn't a raid edition, which doesn't mean we don't want to fix it, but rather we won't get help from the drive manufacturer in trying to troubleshoot it with a raid controller."

[odditory]

Now I am testing Samsung f4 2tb in raid-5. So far so good Looks like we have a cheap winner for Areca

Not much performanceloss wit raid-5 vs raid-0. (read)
Firs a small array:

I'm not ready to declare Samsung F4 a cheap winner for Areca just yet, I have 4 and want to buy more since the Hitachi 2Tb's have all dried up and they haven't released the new 6Gbps gen yet, and before I buy anymore F4's I want to determine if the data loss issue (link below) isn't one that affects these drives on an Areca in RAID or JBOD mode. I only heard about this yesterday so I haven't had time to determine if this is just FUD or a legitimate issue that affects more than just the hardware combos they tested:

http://sourceforge.net/apps/trac/sma...gF4EGBadBlocks

[Rubycon]

I had this issue with 1680ix-24s that were running arrays with 6-16+ SAS 15K drives (Fujitsu MBA series). Only if the system were suddenly powered off, sometimes a blue screen and subsequent shutdown would cause it.

Even if the power was disconnected from the mains a faint beeping every few seconds could be heard. If the machine was restarted it would hang after mainboard POST.

Removing the ribbon cable on the BBU would restore operation.

Have not seen this with our 1880s yet however 90% of our testing has been done WITHOUT the BBU.

[Sandon]

That sounds like the standard behavior of areca controllers when they have disk cache which was not written to the disks. Of course the hanging after boot is not normal.

When it is 'Waiting for Firmware' it will typically write disk cache to the drives which is why it can sit on this screen longer than normal (I have seen disks show major disk activity for 5+ seconds while this happens).

Its very rare that I see it totally hang on that unless there is a problem with one of the drive in the array and the controller is unable to write the data to the drive during array initialization. The few times I have seen this one of the drive had to be pulled shortly after due to errors.

[Rubycon]

That sounds like the standard behavior of areca controllers when they have disk cache which was not written to the disks. Of course the hanging after boot is not normal.

When it is 'Waiting for Firmware' it will typically write disk cache to the drives which is why it can sit on this screen longer than normal (I have seen disks show major disk activity for 5+ seconds while this happens).

Its very rare that I see it totally hang on that unless there is a problem with one of the drive in the array and the controller is unable to write the data to the drive during array initialization. The few times I have seen this one of the drive had to be pulled shortly after due to errors.

I got video of it happening last January here.

http://www.youtube.com/watch?v=jaoIDEnbR0o

[Siobhan]

Hi everyone. I have just recently ordered an 1880ix-24 with the standard 1GB from Provantage (best price @ $1120 at the time). I've read through this thread, and it seems like no one's been able to source a 4GB module that will work reliably a-la-cheap, i.e. under $150. I see that Areca itself has a SKU for it's own certified memory module (as shown here), but that's $253 as of this post. I'd like to upgrade, but I'll be damned if I spend $250 for what is, at best, $110 worth of memory -- registered, ECC or otherwise... So, what I also thought was peculiar while reading the SKU description, is that it says it is "for Arc-1880IX Series Only, 4GB, 240P, PC2-6400, CL5,36C, 256X4, Registered ECC, DDR2"... does anyone notice a few things that pop out right away? First, is the 36C, the next is that it's described as 256x4 -- not 256x8! So that gets me thinking -- will it really accept 256x4 in a 36 chip config? maybe even 512x4 in an 18 chip config? Has anyone actually tried a few of those variables out, or found any reliable inexpensive source of memory that will work, or spoke to Areca on the subject of RAM for the 1880 series?

Thanks in advance for your input.

[SteveRo]

Mr. Computurd has spoken to Areca.
It is a difficult problem.
I had to return 2 sets of RAM that i thought would work and did not.
Other folks also had to return sticks as well.
I ended up spending the $ for the Areca memory - works finebut a shame the price is so high.

[Siobhan]

Thanks for that insight, Steve.

Is the general consensus such that it is not worth playing with ordering RAM that might work, only to find out it doesn't, and to just bite the bullet and go with Areca's own endorsed SKU? I really was hoping someone would have had luck with another (less expensive) brand. It's too bad there are those unresolved, unanswered questions about the Mushkin part.

[Computurd]

yes, the ram issue is a tough one. the kingston 2gb stick is a good value, and it works i have one. the kingston 4gb stick is discontinued from manufacturing, but you can find them for 400 dollars usd.
the unigen ram works beautifully, and you might just have to bite the bullet on it. i rmad tons of sticks trying to find a cheaper way out, but no dice.
tbh 250 isnt that bad for it. the mushkin will not work if memory serves correctly, but i did not test it personally.

[Rubycon]

Honestly I would not mess around trying to find a module that works!
Unigen is certified and works perfectly. Even if you find one that seems to work you may find out later that you have issues. At least 1bit DRAM (ECC) errata will make the controller beep.

4GB is a great amount of information to be trusting to a buffer that may not be solid. If it's for play and it matters not if you hose your array/data contents, etc. and restoring an image is fine then I suppose "playing around" with other modules is ok.

[Siobhan]

By Unigen, do you all mean the "ARC-3404GB" part from Areca as sold on Provantage?

[Rubycon]

By Unigen, do you all mean the "ARC-3404GB" part from Areca as sold on Provantage?

Yes that looks like the same one here:

http://www.pc-pitstop.com/sas_contro...ARCRMD24G8.asp

[SteveRo]

I tried the mushkin (proline) memory - it did not work in my areca 1880.
Had to return it.

[Computurd]

new firmware!

ftp://ftp.areca.com.tw/RaidCards/BIO.../149-20101210/

[Wintel64]

new firmware!

ftp://ftp.areca.com.tw/RaidCards/BIO.../149-20101210/

Can't find the Firmware Release Note yet.

[Nizzen]

Release Notes? What is that Areca? :p

[Sandon]

Can't find the Firmware Release Note yet.

The biggest reason for me to upgrade to this is Finally the 500 ARC-1222 controllers that are used at work can have disks failed remotely. When a disk keeps getting read/timeout errors and the controller doesn't kick it and its causing issues to the point of the machine just about locking up I no longer have to manually pull the drive.

This is a very important feature due to the fact that around 300 of our areca servers are in a data center around 16 miles from our primary one which we do not normally have on-site staff. I asked areca to implement this a while ago so I am really happy they did even though they did not explicity mention it in the changelog:



Changelog:

********* Change Log For V1.49 Firmware

2010-1-8
1 Improve SMP support for ARC1680
2010-1-27
1 ARC1680/1212/1682 4.9.2.0 transport firmware N.G. for some HDDs
use 4.7.3.0
2010-2-4
1 Fix VOLUME_FAILED cause controller to hang
2010-4-8
1 ARC1880:1st release
2010-4-12
1 Add:HP SAS EXP Card 2.02 support
2010-4-15
1 SAS_HOST_MODE:support 6G expander
ARC1680 ....
ARC1880/8040 ....
2 ARC1880/8040:change STACK_SIZE_LOW to avoid problem when LCD panel or VT100
operate HDD fucntion
3 ARC1880/8040:Add LCD initialization strings (test SDRAM ...)
2010-4-16
1 LCD panel:remove checking of MAX 3 hot spare
2010-4-19
1 CheckMigratingX:added for migration failed but migration status is
inconsistent
2 Remove inconsistent fail flag for raid6 migration
2010-4-29
1 Improve checking of spurious SCB/IO, move the check to LinkErrorDevHandle
2 Fix removing of failed HDD hang the system
2010-4-30
1 ARC1880/8040:Fix LSI6G PLI implementation of Start/StopUnit and SATA
passthrough problem
2010-5-5
1 SUPPORT_CHANGE_CAPACITY:after volume capacity is expanded, return check
condition with CAPACITY DATA Changed
2010-5-7
1 ARC1880/8040:RaidInitZero may break, because raid may be freed before
init routine to complete
2 During volume check of RAID4/5/6, hot removing of entire raidset will cause
raidNumberOfDevices to 0, and checking routine will hang
2010-5-12
1 Fix when expasion raidset and target disk failed, after migration pasued
and controller restarted, the migration halted
2010-5-14
1 Add DISK_WRITE_CACHE_SETTING for models without BBU to control disk write
cache mode
2010-5-17
1 Fix race condition for cache line allocation
2010-5-18
1 Fix SAS/SAS2 model's SPC3 compliance
2010-6-1
1 Add ThreadDelayMs to improve GUI speed of PCI host adapter
2010-6-4
1 Tune ARC1880 performance
2010-6-28
1 Fix ARC1880:1st level attached device's RED LED not on if device removed
2010-6-29
1 Enclosure.c:Add space between E%d and %s
2010-7-2
1 ARC1880:fix zoning problem
2010-7-5
1 ARC1880/1680:Add 6G expander zoning support
2010-7-20
1 Fix TEST_ZONING with log
2 Fix enclousre power off with raidset is initializing or rebuilding...
2010-7-27
1 ARC1880:modify geteon delay stepping,
2010-7-28
1 Add support for mail greet pause, use different tcp_send timeout value
2010-7-29
1 Fix raidset activate problem for rebuilding percentage
2010-7-30
1 Add display of Error Recovery Control Status of SATA hdd
2010-8-19
1 Modify pure SATA's NEW_83782D_HWMON SNMP
2010-8-20
1 Patch ARC1880 bootcode for PLI Phase6/7
2010-8-30
1 Add CheckEesaBlackList to exclude some HDD with error recovery control
problem
2010-9-2
1 Add SUPPORT for ARC1880IXL-8
2010-9-9
1 ARC1880:fix return from BBM with 4G sdram
2010-9-16
1 Fix when many volumes is created for RAID6 and failed two drive, some volume
may still NEED_REBUILD and REBUILDING
2010-9-17
1 Fix SAS devices add/remove routine is invoked before SasDeviceInit is called
2010-10-6
1 Add support for ARC1880LP
2010-10-7
1 Modify GetResource and PostResource
2010-10-15
1 Fix write through problem for IOP348 based HBA
(0) ARC1680/1212/1682/SPT1680/WTN1212
(A) Only ARC1680 reveal the problem
(B) The problme occurred if SUPPORT_PERSISTENT_RESERVE is defined
(C) The problem is the STACK overflow if write through
(D) Change stack size for All IOP348 models
2010-10-20
1 Fix reportlun command to exclude FAILED and FOREGROUND init volume
2010-10-21
1 ARC1680:Add LARGE_DEV_LBA support
2010-10-22
1 SAS2 target mode:ARC8040/8066/8366
2010-10-28
1 Fix MRVL5182 based more than 2TB hdd support
2010-11-3
1 ARC1682:Add over 2TB hdd support
2010-11-5
1 ARC1680:remove VerifyDrive to support seagate SED drive
2010-11-25
1 Add NO_UPS_STATUS for enclosure.c to remove ENC without UPS status
2 Patch supermicro SES problem, unsupported power state report value 6
2010-11-30
1 Add rename raidset name support
2 Port ARC1880 subsystem
4 Add NEW_MAIL_ALERT_CONFIG to support different mail alert config for
different mail account
2010-12-2
0 V1.49 20101202
1 All PCI/E HBA models (except 1880)
2010-12-10
1 V1.49 20101210
2 ARC1880

[Nizzen]

Mailed Areca. got answer from 2 different people in areca:

1. Is there a new FW? I did not know about that

2. There is a new FW, it was made after a client request

[Computurd]

LOL nizzen, Areca does seem to be very vague about their FW updates...no big changlelogs etc...i havent really tested the new version very much, been very busy. but i do have it on here and it seems to be working very well...