Support for MaprFS

The filesystems are not mounted but show up using iostat. Please make it visible in instana. Reference for MaprFS (https://docs.datafabric.hpe.com/60/MapROverview/c_maprfs.html)

Idea priority

Low

Post comment

Guest

Reply | Jun 30, 2023

Hello, I am part of the group that requested this feature. Let me explain the request.

First, let me say that this behavior is not specific to MapR, but we are seeing it with MapR. Minio is another example of a software that works like this.

MapR uses the disks in the systems directly, as a block device. There is no mounted file system on the disks that MapR uses.

$ lsblk
NAME                          MAJ:MIN RM   SIZE RO TYPE MOUNTPOINT
sda                             8:0    0   100G  0 disk
├─sda1                          8:1    0     1G  0 part /boot/efi
├─sda2                          8:2    0     2G  0 part /boot
└─sda3                          8:3    0    97G  0 part /
sdb                             8:16   0   3.7T  0 disk
sdc                             8:32   0   3.7T  0 disk
sdd                             8:48   0   3.7T  0 disk
sde                             8:64   0   3.7T  0 disk

As seen, sda is the system disk, while sd[b-e] have no file system. These are the disks that are used by MapR. As list of disks that is used by MapR is available at /opt/mapr/conf/disktab.

$ cat /opt/mapr/conf/disktab
# MapR Disks Thu Mar  2 17:27:49 2023

/dev/sdb 1a2aa006-6847-467e-816f-94dfc309adff
/dev/sdd 046aac68-bc74-4982-ac6e-b6ca3e791076
/dev/sde 310ea197-8954-4230-a027-bf8cc7434ef6
/dev/sdc b5b5fc49-0a54-47f0-97cf-76c4ae234deb

Regarding stats that are important to us, the tool sar gives a good starting point.

$ sar -p -d 1 1
Linux REDACTED (REDACTED)     06/30/2023      _x86_64_        (REDACTED CPU)

12:59:47 PM       DEV       tps     rkB/s     wkB/s   areq-sz    aqu-sz     await     svctm     %util
12:59:48 PM       sda      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
12:59:48 PM       sdc      2.00     24.00      0.00     12.00      0.04     17.50     19.00      3.80
12:59:48 PM       sdb      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
12:59:48 PM       sdd      0.00      0.00      0.00      0.00      0.00      0.00      0.00      0.00
12:59:48 PM       sde      1.00      8.00      0.00      8.00      0.02     20.00     21.00      2.10
[...]

The important stats are utilization in percentage and await (the average time for I/O requests issued to the device to be served).

The goal is to find highly utilized disks or disks with a high latency within the cluster.

It may or may not make sense to capture the processes that generates the I/O, as this should always be the executable /opt/mapr/server/mfs.

Please let me know if there is any further information that is required to implement I/O monitoring of non-mounted disks.

Kind regards,

Markus

Admin

Rodoljub Radivojevic

Reply
| Jun 2, 2023

This feature has been planned for 4Q 2023.

0 reply Hide replies

0 MERGED