[LU-18041] struct obd_uuid is too small for MGC"NID" uuid with IPv6

• Type: Bug
• Resolution: Unresolved
• Priority: Critical
• Fix Version/s: Lustre 2.18.0
• Affects Version/s: Lustre 2.16.0, Lustre 2.17.0
• Labels: IPv6 tstu_ipv6
• Environment: MGS configured with IPv6 strings that can't be compressed into UUID_MAX size.
• Severity: 3
• Assignee: James A Simmons
• Reporter: James A Simmons

Description

IPv6 NIDs can vary in size, up to 42 bytes in total when expressed in ASCII and the leading MGC is added to the MGC name, which is slightly larger than struct obd_uuid maximum size UUID_MAX = 40 bytes:
https://testing.whamcloud.com/test_sets/4ae697a2-d2db-4791-b14e-014bb6e67487

Lustre: DEBUG MARKER: mount -t lustre -o user_xattr,flock fd33:3981:3213:f010:0:5254:14:c527@tcp:/lustre /mnt/lustre
LustreError: 16630:0:(mgc_request.c:1456:mgc_apply_recover_logs()) mgc: cannot find UUID by nid 'fd33:3981:3213:f010:0:5254:eb:f685@tcp': rc = -2
Lustre: 16630:0:(mgc_request.c:1635:mgc_process_recover_log()) MGCfd33:3981:3213:f010:0:5254:14:c527@tcp: error processing lustre-cliir log recovery: rc = -2
Lustre: 16630:0:(mgc_request.c:1899:mgc_process_log()) MGCfd33:3981:3213:f010:0:5254:14:c527@tcp: IR log lustre-cliir failed, not fatal: rc = -2
LustreError: lustre-MDT0000-mdc-ff12b8ef43b85800: operation mds_connect to node fd33:3981:3213:f010:0:5254:14:c527@tcp failed: rc = -11

Update places in the kernel where obd_uuid is used for MGS. 

Issue Links

• is duplicated by: LU-20043 lustre-initialization IPv6: mgc_apply_recover_logs() mgc: cannot find UUID by nid 'fd33:3981:3213:f020:0:5254:c8:10ad@tcp'
• is duplicated by: LU-20125 lustre-initialization: timeout on mount - mgc: cannot find UUID by nid
• is related to: LU-18210 sanityn test 113: fails with cannot find mgc of nid
• is related to: LU-17594 Simplify ptlrpc_connection handling
• is related to: LU-18417 Finish IPv6 support
• is related to: LU-10359 remove NIDs from config llogs
• is related to: LU-13118 change client instance to respect ASLR
• is related to: LU-18884 make LWP setup based on IR instead of client config
• is related to: LU-19360 lustre-initialization: IPv6 "Failed to get NID for server"
• is related to: LU-10391 LNET: Support IPv6