0

Due to some changes, I have to force a rescan of Fibre Channel devices on a CentOS 6 server.

This is the output of multipath -l:

(...) 36000144000000010f01c857894aede59 dm-50 EMC,Invista size=5.0T features='1 queue_if_no_path' hwhandler='0' wp=rw `-+- policy='round-robin 0' prio=0 status=active |- 7:0:3:107 sdca 68:224 active undef unknown |- 4:0:3:107 sdbd 67:112 active undef unknown |- 7:0:4:107 sdeb 128:48 active undef unknown `- 4:0:7:107 sddg 70:224 active undef unknown 3600601602bd14600351eb55f237aa77d dm-5 DGC,VRAID size=3.0T features='0' hwhandler='1 emc' wp=rw |-+- policy='round-robin 0' prio=0 status=active | |- 7:0:7:39 sdq 65:0 failed undef unknown | `- 4:0:6:39 sdei 128:160 failed undef unknown `-+- policy='round-robin 0' prio=0 status=enabled |- 7:0:0:39 sdx 65:112 active undef unknown `- 4:0:0:39 sdk 8:160 active undef unknown 36000144000000010f01c857894aedd26 dm-14 EMC,Invista size=5.0T features='0' hwhandler='0' wp=rw `-+- policy='round-robin 0' prio=0 status=active |- 7:0:3:94 sdba 67:64 failed undef unknown |- 4:0:3:94 sdad 65:208 failed undef unknown |- 7:0:4:94 sddb 70:144 failed undef unknown `- 4:0:7:94 sdcg 69:64 failed undef unknown (...)

Running multipath -f dm-5 or multipath -w 3600601602bd14600351eb55f237aa77d does not delete the entry, not even after running multipath or restarting the multipathd service.

These entries are not present in /etc/multipath.conf.

How to delete the failed entries?

Improve this question
1

2 Answers 2

Reset to default
1

I have to force a rescan of Fibre Channel devices on a CentOS 6 server.

You can do that by the (rather unintuitive):

echo "- - -" > /sys/class/scsi_host/<host>/scan

Replace "<host>" with the respective FC-host.

As a side note: you may want to use more but smaller LUNs if performance is an issue. THe reason is that for a every disk a command queue is maintained and this can only hold so many entries. So, especially if you have many small disk operations (typical for e.g. databases) you may want to use rather several 1TB- or 2TB-LUNs instead of the single 5TB LUNs your system uses.

Improve this answer
1
  • Running multipath -f <device> and then for h in /sys/class/scsi_host/host*; do echo "- - -" > /sys/class/scsi_host/$h/scan; done solved my issue. Commented Nov 4, 2024 at 9:54
1

steps for deleting 36000144000000010f01c857894aedd26 dm-14:

multipath -f 36000144000000010f01c857894aedd26 blockdev --flushbufs /dev/sdba blockdev --flushbufs /dev/sdad blockdev --flushbufs /dev/sddb blockdev --flushbufs /dev/sdcg echo 1 > /sys/block/sdba/device/delete echo 1 > /sys/block/sdad/device/delete echo 1 > /sys/block/sddb/device/delete echo 1 > /sys/block/sdcg/device/delete
Improve this answer

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.