Issue with Adapec RAID not pulling all S.M.A.R.T. stats

[mcesnik]

So this isn't so much a problem with smartctl as it is with Adaptec. I am creating this issue instead of a PR because the solution is a bit of a hack and I just want to provide insight and leave it at that.

NB - Since this solution uses Python (as I do not know Go) you will need to create your own Dockerfile.

First I need to create a wrapper to smartctl that would go into /scrutiny/bin. You also need the arcconf binary to be in the /scrutiny/bin.

#!/usr/bin/python3

import os
import sys
import json
import xml.etree.ElementTree as ET

arcconf = "/scrutiny/bin/arcconf getsmartstats 1"
smartctl = "/usr/sbin/smartctl"

smart_stats = {
    1: "Raw_Read_Error_Rate",
    3: "Spin_Up_Time",
    4: "Start_Stop_Count",
    5: "Reallocated_Sector_Ct",
    7: "Seek_Error_Rate",
    9: "Power_On_Hours",
    10: "Spin_Retry_Count",
    11: "Calibration_Retry_Count",
    12: "Power_Cycle_Count",
    192: "Power-Off_Retract_Count",
    193: "Load_Cycle_Count",
    194: "Temperature_Celsius",
    196: "Reallocated_Event_Count",
    197: "Current_Pending_Sector",
    198: "Offline_Uncorrectable",
    199: "UDMA_CRC_Error_Count",
    200: "Multi_Zone_Error_Rate"
}

if __name__ == "__main__":
    # Run smartctl and load the Json into memory
    cmd = "%s %s" % (smartctl, " ".join(sys.argv[1:]))
    output = json.loads("".join( [ i.strip() for i in os.popen(cmd).readlines() ] ))

    # Check to see if this device is actually for aacraid and do this 'extra' work if so
    aacraid = [arg for arg in sys.argv if arg.startswith("aac")]

    if any(aacraid):
        # Get the device id 
        id = int(aacraid[0].split(",")[3])

        # Call arcconf getsmartstats for controller 1 and then get the XML we care about...
        stats = [dev for dev in ET.fromstring("".join([i.strip() for i in os.popen(arcconf).readlines() if i[0] == "<"][0:-2])) if int(dev.attrib["id"]) == id ][0]

        # Now get the attributes for the device we extracted
        attributes = [ (int(stat.attrib["id"], 16), stat.attrib["normalizedCurrent"], stat.attrib["normalizedWorst"], stat.attrib["rawValue"]) for stat in stats ]

        # Write the ata_smart_attributes (I have no idea if these matter for a SCSI or not)
        output["ata_smart_attributes"] = {
          "revision": 16,
          "table": [ {
                      "id": attribute[0],
                      "name": smart_stats[attribute[0]],
                      "current": int(attribute[1]),
                      "worst": int(attribute[2]),
                      "raw": { "value": int(attribute[3]), "string": attribute[3] }
                     } for attribute in attributes if attribute[0] in smart_stats ]
        }

        # And write these additional properties too (scrutiny uses these for sure)
        output["temperature"] = { "current": [ int(a[3]) for a in attributes if a[0] == 194 ][0] }
        output["power_on_time"] = { "hours": [ int(a[3]) for a in attributes if a[0] == 9 ][0] }
        output["power_cycle_count"] = [ int(a[3]) for a in attributes if a[0] == 12 ][0]

    # Dump the whole thing to stdout just like smartctl
    print(json.dumps(output, indent=4))

For reference, this my Dockerfile

FROM analogj/scrutiny

EXPOSE 8080

RUN apt -y update && apt -y upgrade
RUN apt -y install tzdata python3

COPY arcconf /scrutiny/bin
RUN chmod 755 /scrutiny/bin/arcconf

COPY smartctl /scrutiny/bin
RUN chmod 755 /scrutiny/bin/smartctl

After building a new image using docker build you can create a container as per the standard documentation but before you do add the following to a collector.yaml and include that in the /scrutiny/config as a volume mount.

devices:
  - device: /dev/sd-
    type:
      - aacraid,0,0,0
      - aacraid,0,0,1
      - ... the rest of your drives here

I hope this helps someone save some time if they are using some legacy Adaptec cards like me :-)

[mcesnik]

Also, I noticed that although $TZ is supposed to be supported the tzdata package is not installed.

[AnalogJ]

@mcesnik Appreciate your patience and thanks for the helpful python script.

The beta branch (and beta-* images) support a new collector config file option: commands.metrics_smartctl_bin
which can be use to customize the smartctl binary path.

I'll also add this to the documentation section.

Thanks again!

[AnalogJ]

though, I am curious why is this necessary? smartctl is supposed to support Adaptec raid devices natively. Whats the output of commands like: smartctl -a -d "aacraid,0,0,1" /dev/sda