E-Series Infiniband SRP hosts lose access to storage array
Applies to
- NetApp E-Series
- NetApp EF-Series
- InfiniBand host protocol
Issue
Infiniband SRP rejections occur causing hosts to lose access to storage array.
- DQ logs:
10/02/21-20:42:32.062536 01 lclWorker1_0 srp cffff Host login REQ, single: True I_T_IU_Len: 260 srpOpcode: SRP_LOGIN_REQ tag: 0x0
10/02/21-20:42:32.062537 01 lclWorker1_0 srp cffff Initiator Port Id: 0c 42 a1 03 00 f9 ae 00 e4 1d 2d 03 00 26 48 72
10/02/21-20:42:32.062538 01 lclWorker1_0 srp cffff Target Port Id: 24 13 d0 39 ea 42 26 9c d0 39 ea 03 00 f9 ae 00
10/02/21-20:42:32.062548 01 lclWorker1_0 srp cffff Failed to allocate RDMA channel, cm_id: 0x346384560
10/02/21-20:42:32.062551 01 lclWorker1_0 srp cffff HOST LOGIN REJECT, tag: 0x0 status: 0
- Host Logs:
2021-10-02T20:42:32.029837-04:00 gpfsbal811 kernel: [40024.793161] scsi host50: ib_srp: Connection 0/8 to fe80:0000:0000:0000:0c42:a103:00f9:af28 failed
2021-10-02T20:42:32.393831-04:00 gpfsbal811 kernel: [40025.157781] scsi host49: ib_srp: REJ received
2021-10-02T20:42:32.393834-04:00 gpfsbal811 kernel: [40025.157782] scsi host49: REJ reason 0x3