LSI MegaRAID

---

Install

The MegaRAID software installs /usr/local but we actually want it in /opt/services. So we make a symlink.

To install the software, be root

cd /tmp
tar xzvf /home/src/lsi/MSM_linux_x64_installer/current.tgz
cd /tmp/disk
./install.csh -l
ln -s "/usr/local/MegaRAID Storage Manager" "/opt/services/MegaRAID"

You may see errors about systemctl. Systemctl is an RHEL7 feature and it can be safely ignored.

Check that the daemon is running

/etc/init.d/vivaldiframeworkd status

If it isn't, start it with

/etc/init.d/vivaldiframeworkd start

Cleanup

cd /
rm -rf /tmp/disk

---

Configure

Consistency Check (cc) and Patrol Read (pr)

The cc and pr can take days to complete. The first Virtual Drives on our nmngas nodes have a strip size of 128KB while the other Virtual Drives have a strip size of 256KB. This may be what causes the cc to take twice as long to finish on VD_0 than VD_1 or VD_2. As of Apr. 20, 2016, VD_0 takes about 84 hours to finish a cc while VD_1 and VD_2 take about 41 hours. This means we need to sufficiently stagger cc and pr like on alternating weekends.

Consistency Check (cc)

cc reads all portions of a stripe, computes the parity and compares this against the parity on disk. If the compare fails, the correct parity is written on disk. This is a more logical check than pr.

Patrol Read (pr)

pr reads all sectors of all disks, including hot spares. If it finds an error, steps are taken depending on the type of array. This is a more physical check than cc.

How to see what the cc and pr are set to

/opt/services/MegaRAID/StorCLI/storcli64 /call show cc
/opt/services/MegaRAID/StorCLI/storcli64 /call/vall show cc
/opt/services/MegaRAID/StorCLI/storcli64 /call show pr

Or set cc and pr to run on alternate weekends starting Friday night at 11pm. The syntax requires an actual first date to start the process and that date be in the future

cc_start="`date '+%Y/%m/%d' --date='friday'` 23"
/opt/services/MegaRAID/StorCLI/storcli64 /call set cc=conc starttime=${cc_start} delay=336
pr_start="`date '+%Y/%m/%d' --date='friday 7 days'` 23"
/opt/services/MegaRAID/StorCLI/storcli64 /call set pr=on mode=auto starttime=${pr_start} delay=336

Drift Issue

#krowe Aug 4 2016: some of the OSS nodes drift when they start the CC and PR. The two processess seem to drift the same and it has so far always been slow by a day or two (i.e. starting on Thursday instead of Friday). #krowe Aug 3 2016: Probably because I set a delay of 336 hours. But I bet it starts the next cc or pr 336 hours after it finishes the current one and not after it starts. Hence the drift. Grrr.

---

Use

Follow the VNC instructions to get a graphical login.

To launch the User Interface (don't be root)

/opt/services/MegaRAID/startupui.sh

The software looks at the local password file so you need to login as root with root's password for that machine.

Usefull command lines

Show Controller (ROC) Temps

/opt/services/MegaRAID/StorCLI/storcli64 adpallinfo aall | grep -i temp

Show Disk Temps

/opt/services/MegaRAID/StorCLI/storcli64 /c0/eall/sall show all | grep -i temp

Show Cache Vault (battery backup) Temp

/opt/services/MegaRAID/StorCLI/storcli64 /call/cv show status | grep -i temp

Show status of all cards and enclosures

/opt/services/MegaRAID/StorCLI/storcli64 /call/eall show status

Show all disks on card 0

/opt/services/MegaRAID/StorCLI/storcli64 /c0/dall show

Show warning events of card 0

/opt/services/MegaRAID/StorCLI/storcli64 /c0 show events type=sincereboot filter=warning

Replace a failed drive

Description	MegaRAID Storage Manager	storcli64	Example storcli64
1 List drives	Use Physical and Logical tabs	/cx show
2 Locate the failed drive	Right-click and select Start Locating Drive	/cx[/ex]/sx start locate	/opt/services/MegaRAID/StorCLI/storcli64 /c0 /e62 /s1 start locate
3 Physically locate the drive and label it with a sticky note
4 Stop locating drive	Right-click and select Stop Locating Drive	/cx[/ex]/sx stop locate	/opt/services/MegaRAID/StorCLI/storcli64 /c0 /e62 /s1 stop locate
5 Make drive offline or UBad	Right-click Make Drive Offline	/cx[/ex]/sx set offline	/opt/services/MegaRAID/StorCLI/storcli64 /c0 /e62 /s1 set offline
6 Prepare drive for removal	Right-click Prepare for Removal	/cx[/ex]/sx spindown	device state doesn't support requested command
7 Physically remove the drive
8 Physically insert a new drive
9 List drives	Use Physical and Logical tabs	/cx show	(RAID 6 will rebuild) 62:1 39 Rbld 0 3.637 TB SATA HDD N N 512B WDC WD4000FYY U
10 Set new drive to UGood (not necessary for RAID 6)	Right-click and select Unconfigured Good	/cx[/ex]/sx set good	(only on RAID 5)
11 Set new drive to DHS (not necessary for RAID 6)	Right-click and select Dedicated Hot Spare	/cx[/ex]/sx add hostsparedrive dgs=x	(only on RAID 5)
12 Ensure the alarm is off	Click on Silence alarm	/cx set alarm=silence	/opt/services/MegaRAID/StorCLI/storcli64 /c0 set alarm=silence

---

Resources

• to do: provide links to documents for host bus adapters (HBA) in the storage servers
• James's notes for creating arrays with the GUI interface

---