SAN’s

The first thing to do when troubleshooting anything is to gather information. The information that should be gathered when troubleshooting possible SAN problems include(but should not be limited to) the following:

  1. How is it used? Virtualization Datastore, Windows iscsi target, NAS?
  2. Is data on the SAN inaccessible, or just slow?
  3. SAN IP address information (Be it Virtual IP for Lefthands, or Management IP)
  4. What SAN Hardware vendor, make and model are we working with?
  5. What type of fabric does it connect with? (Fiber Channel, iSCSI, FCoE, other)
  6. SAN fabric vendor, make, and model?
  7. Has anything changed recently? Has anything changed since implementation?
  8. Are the IP address(s) pingable from a workstation? (Should they be?… sometimes the SAN is segregated and isn’t pingable during normal operation either)

Next, after gathering information, start to determine if you really have a SAN problem, because SAN’s usually rely on some sort of fabric switch for connectivity, lets check that first.

  • Make sure that the SAN is powered on, and most importantly has no lights that are indicating failure. If there are indicator lights, consult the vendor and documentation to troubleshoot those lights.
  • Make sure that the fabric switches have power and have no amber or red lights (or ports not lit up that should be) … pretty much anything but green = bad.
  • Check cables between switch and SAN ports, make sure there are link lights if equipped.
  • If the switches are fully functional next check to see if all services provided by the SAN are down, it could be possible that only certain servers have lost connectivity.
  • If only one or a few servers have lost connectivity to the SAN check cables,  after checking cables going from the switches to the servers, make sure that link lights are present on the HBA’s as well as the switch ports.
  • If all servers have no SAN connectivity then we know that its either the cables between the switch and the SAN, a SAN hardware, or a SAN software issue. We assume that multiple servers are not having the same problems at the same time, it is more likely that there is a problem with the san.

If there are no visible hardware issues (like cables unplugged, or power issues), we need to move on to more vendor specific troubleshooting.

Disk alignment. The following powershell command will check a windows machine to see if its file systems are aligned properly or not.
Get-WmiObject Win32_DiskPartition | FT Name, @{Label="Align"; Expression={if($_.StartingOffset%4096 -eq 0){$True} else { $False}}} -auto
If you add a -ComputerName “ComputerName” behind Win32_DiskPartition you can also run it remotely on another computer.

2 Responses to "SAN’s"

  1. Regarding disk alignment what are the consequences if the script returns false? Like this example below:

    Name Align
    —- —–
    Disk #0, Partition #0 False
    Disk #0, Partition #1 False
    Disk #1, Partition #0 False

    Does it affect performance of this VM and what can be done to fix this?

  2. IT would affect performance of the VM. Basically if a block of data at the virtual machine level is spread into 2 blocks at the SAN level, then the SAN has to do two reads or two writes for every 1 write (or read) that the VM requests.

    Search google for disk alignment tools. Most of the time you would want to do this BEFORE installing the OS, but if you already have data on it then there are still some tools out there that could rearrange the data, but downtime is almost guaranteed while it does it.

Post Comment