We have two C6220 (not II) with LSI 9265-8i (one filled with 2*E5-2680 with 128Gb of RAM on each nodes and the other with 2*E5-2670 with 96Gb of RAM each) that we bought recently and we are experiencing huge issues with SATA SSDs : Disks drops out of RAID (sometimes 3 or 4 disks at once which can be catastrophic...).
Its easier to trigger these drops by doing a reboot or when there is an important load on the RAID but sometimes it just drops one or many disks at once even when on the LSI Webbios not doing any reconstructions or so...
It seems that some SSDs are less prone to be dropped than others and some of the enclosure slots are also less prone to drop disks than others :
- We didnt had any issues so far (about 2 months) with 6*Sandisk Extreme Pro 960Gb on the #1 sled of our first C6220
- No issues no far (1 week) with 4*Samsung 850 Pro 480Gb on the #4 sled of our first C6220 (but there were drops on the #2 and #3 sleds of the same chassis...)
- Tested Dell branded Seagate Savvio 10k.6 300Gb SAS disks on the second chassis for more than a week without any issue so far
- Tested 4 and 6*Sandisk Extreme Pro 480Gb on the sleds #2, #3 and #4 of the first chassis and it was experiencing drops in all cases (even more in #2 and #3 where is couldnt even boot at all with the 6 disks detected at the same time)
- Tested with the Sandisk Extreme Pro 480Gb on the other C6220, we had issues as well
- Tested 2, 4 and 6*Samsung 850 Evo 250Gb on the second C6220 on every sleds and its dropping regularily
Here is what we tried so far without luck :
- Both 2.5.3 and 2.7.1 BIOS (C6220 II bios is compatible with C6220 even if not stated clearly on the website)
- The LSI card initially had firmwares from 2013, we upgraded to the latest one without any difference
- Forcing the link speed to 3Gbps instead of the default 6Gbps
- Exchanging sleds where enclosure slots seems more prone to drops with another one which was behaving better... the behavior didnt change so the problem probably midplane or backplane related
- Verified that all cables were correctly plugged, unplugged and replugged them
On the RAID card log i have many of these :
Event Description: Unexpected sense: PD 12(e0xfc/s1) Path 4433221102000000, CDB: 28 00 00 19 d6 00 00 02 00 00, Sense: b/47/03
Then when a disk disconnects :
Event Description: PD 12(e0xfc/s1) Path 4433221102000000 reset (Type 03)
Event Description: Removed: PD 12(e0xfc/s1)
Event Description: Removed: PD 12(e0xfc/s1) Info: enclPd=fc, scsiType=0, portMap=02, sasAddr=4433221102000000,0000000000000000
I found an old topic with similar issues with 840 Pro but it had been solved with firmware upgrades if i understood correctly :
Should be sticky: Samsung 840 and 840 pro are not LSI megaraid compatible
Does anyone had any issues such as this one?
Buying Dell SSD is not a viable solution as their price is way too high, the servers seller tells us that he can sell us these SSD models that should be compatible while not too expensive : "Intel 335 SSDSC2CT240A4" and "Adata XPG SX900 ASX900S7-256G".
Do you think that its a viable solution?