When I reconfigured my server and swapped my smaller SAS drives for larger SATA drives into my NetApp DS4246 all of the sudden the log is being spammed constantly every couple of minutes with this exact error:
I never had this problem with my SAS drives and the only other thing I changed with the setup was going to a LSI 9305-16e from a 9300-16i (with 8i8e device to connect to the shelf), however trying to go back to the 9300 as well as a 9206-16e I had before that one still result in the same spamming of the log with the same error. All three HBAs are on the latest firmware but Iāve tried downgrading them a version or two and also trying ACM and non ACM firmware but every combination just acts exactly the same. The SATA drives were also internal to my old case and hooked to the 9300 via sff8643 to 4x SATA cables and also never had any errors or issues like this. This doesnāt seem to cause any performance or integrity issues at least from what I can tell, but Iāve noticed some of the drives tend to load up with a long backlog shown by netdata of 5-10 seconds when under constant load (like large file ops or torrent downloads) which I donāt think happened before either. Has anyone had a problem like this or know how to fix it? Iām tempted to get a supermicro cse846 or 847 in hopes of fixing it, but they arenāt exactly cheap and who knows if the models with an expander backplane (which are much cheaper and more available than single lane per device ones) wonāt just have the same problem.
Not really, itās designed and advertised as such in NetAppās original documentation to work with SAS and SATA without issue, there are interposers for SATA but from what Iāve read online people say to avoid them because they only exist to connect SATA drives to both IOMs to enable multi-pathing, but they cause issues and randomly fail and people say that when they ditched them any problems they had went away.
NetApp may well have supported certain SATA models based on their filesystem and hardware config. My point is youāre running TrueNAS and ZFS on NetApp hardware. All things talking SAS is fairly straightforward but SATA is a second class citizen in this configuration and requires much more specific configuration.
Donāt get me wrong TrueNAS works great with SATA drives when they are directly attached but going via an expander can produce weird and not so wonderful results depending.
I guess, but itās not like enterprise SATA doesnāt exist, seems weird to me that expanders wouldnāt be able to handle them properly, especially ludicrously expensive (when new) business gear built to be as reliable and compatible as possible to keep also ludicrously expensive warranty and contract repair to a minimum.
This is why storage providers often deploy hardware and software as a complete package. What we are doing mostly on this forum is decoupling the two which obviously brings more variables.
The few bug tickets I have seen for items like this get closed due to use of āshelvesā. Not supported under Community Edition. It takes too much resources to support these problems.
i see. guess ill just live with it for the foreseeable future since it (seems to be (mostly)) harmless, until i can afford a supermicro with non expander backplane or something.
well i saw this thread before when first troubleshooting this issue but just skimmed it and ignored it when it didnt have a definitive fix. googling again for an answer i clicked on that thread again and actually read it, the guy said power cycling the shelf after truenas boots fixes the issue at the cost of it being a pain since you have to export and import your pools. well i tried it, and so far for an entire week i have gotten not 1 single mpt3sas error in the log and all of the instability and backlog issues have completely gone away, AND to top it off my sever now uses about 30 watts less at the wall, which is an almost 10% savings. so its not ideal, but it does work.
