Troubleshooting ZFS – Common Issues and How to Fix Them

Summary

ZFS is a very powerful filesystem and volume manager that offers incredible flexibility and resilience. But computers are made up of hardware, and hardware breaks. Like any other component of a computer system, there are times when problems arise that require the attention of the operator to properly resolve. The power and complexity of ZFS can sometimes leave users unsure of what to do when they run into an issue or find themselves needing to resolve an error that they are experiencing. In this article, we'll go over some simple issues that a ZFS user may run into and how to address them. Issue 1: Checksum errors in zpool status pool: zroot state: ONLINE config: NAME STATE READ WRITE CKSUM zroot ONLINE 0 0 0 mirror-1 ONLINE 0 0 0 da2p2 ONLINE 0 0 0 da3p2 ONLINE 0 0 0 mirror-2 ONLINE 0 0 0 da4p2 ONLINE 10 0 24 da5p2 ONLINE 0 0 0 If your monitoring detects an issue, or you run zpool status manually and notice that these error counters are not all 0, the first thought a user might have would be: "Is my data ok?" If you see values in the Read, Write, or Cksum columns, but no warning in the header or footer about unrecoverable files, your data is most likely fine...however, you will need to take a few steps to ensure that the situation does not deteriorate further. Usually when values appear in these columns, it is a sign that one of your disks is starting to fail. Before doing anything, check to make sure you have a recent backup, just in case. Next, check the SMART data for the disk in question and see if it reports anything out of the ordinary. smartctl -x /dev/da4 Take notice of any results for "Current_Pending_Sector” or "Reallocated_Sector_Ct" on SATA disks, or for SAS: “Total uncorrected errors”. Attribute number Attribute number 5: Reallocated_Sector_Count, Type: Pre-fail, Raw value: 80Error Counter Log: Total uncorrectable errors: 0Error Counter Log: Total uncorrectable errors: 0If nothing stands out, run a short SMART test.smartctl -t SHORT /dev/da4Num Test_De...