Hide Forgot
Description of problem: Few existing LUNs are corrupted after doing addition and removal of new LUNs through ceph-iscsi-ansible Version-Release number of selected component (if applicable): ceph-iscsi-ansible-1.5-1.el7test.noarch ceph-iscsi-config-1.5-1.el7test.noarch ceph:- 10.2.3-12.el7cp.x86_64 Steps to Reproduce: 1.configured 30 iscsi LUNs on a 3:1 gateway setup 2.Mounted 15 LUNs on RHEL 7.3 iscsi initiator and ran I/Os on few 3.While I/Os are in progress added 30 more LUNs from ceph-iscsi-ansible 4. On the initiator side newly created LUNs were discovered meanwhile I/O were successfully running on existing LUN mounts 5. Performed removal of the newly added LUNs using ceph-iscsi-ansible Note:- Method of removal is by making status of the client as "absent" in "client_connections" part. ex:- - { client: 'iqn.1994-05.com.redhat:f3ca348ccc5b', image_list: 'rbd.ansible46,rbd.ansible47,rbd.ansible48,rbd.ansible49,rbd.ansible50,rbd.ansible51,rbd.ansible52,rbd.ansible53,rbd.ansible54,rbd.ansible55,rbd.ansible56,rbd.ansible57,rbd.ansible58,rbd.ansible59,rbd.ansible60', chap: 'admin/redhat', status: 'absent' } 6. Performed iscsi-session rescan after the removal Actual results: 1. After the removal some of the existing LUNs are throwing I/O error [ubuntu@magna048 mpathm]$ ls ls: cannot open directory .: Input/output error 2. Some of the mounted LUNs were disappeared after rescanning from dmesg:- ============ [526559.662821] XFS (dm-3): xfs_log_force: error -5 returned. [526589.743377] XFS (dm-3): xfs_log_force: error -5 returned. [526619.823932] XFS (dm-3): xfs_log_force: error -5 returned. [526649.904499] XFS (dm-3): xfs_log_force: error -5 returned. [526679.985053] XFS (dm-3): xfs_log_force: error -5 returned. [526710.065614] XFS (dm-3): xfs_log_force: error -5 returned. [526740.146196] XFS (dm-3): xfs_log_force: error -5 returned. [526770.226732] XFS (dm-3): xfs_log_force: error -5 returned. [526800.307297] XFS (dm-3): xfs_log_force: error -5 returned. [526830.387842] XFS (dm-3): xfs_log_force: error -5 returned. [526835.947415] blk_update_request: 111 callbacks suppressed [526835.952826] blk_update_request: I/O error, dev dm-0, sector 0 [526835.958838] blk_update_request: I/O error, dev dm-0, sector 20971392 [526835.965405] blk_update_request: I/O error, dev dm-0, sector 20971504 [526835.971899] blk_update_request: I/O error, dev dm-0, sector 0 [526835.977763] blk_update_request: I/O error, dev dm-0, sector 8 [526835.983815] blk_update_request: I/O error, dev dm-0, sector 0 [526836.004535] blk_update_request: I/O error, dev dm-2, sector 0 [526836.010475] blk_update_request: I/O error, dev dm-2, sector 62914432 [526836.016964] blk_update_request: I/O error, dev dm-2, sector 62914544 [526836.023521] blk_update_request: I/O error, dev dm-2, sector 0
This is not how LUNs are removed from a client The luns presented to a client are detailed in the image_list variable - to remove luns simply update the image_list by removing the required entries. The status field on the client definition is the status of the WHOLE client definition to the gateways i.e. you only set status=absent to remove the client from the gateway configuration. This is covered in the the deployment guide. Pleas retest using the correct process.
After following paul and mike's suggestion this is not seen, hence marking this as not a bug.
(In reply to shylesh from comment #5) > After following paul and mike's suggestion this is not seen, hence marking > this as not a bug. Lately I realised that LUNs are reshuffled and the LUN's which were properly mounted and not removed are throwing I/O errors. Please refer to the bug https://bugzilla.redhat.com/show_bug.cgi?id=1392378 for more info. Sorry for the inconvience ,reopening the bug.
In this test are you removing a image from the client's image_list or marking the client absent?
The changes look good to me. Mike, Could you take a look at the warning. Thanks, Tejas
Thanks Mike.