nndocs:srp
Differences
This shows you the differences between two versions of the page.
Next revision | Previous revision | ||
nndocs:srp [2025/02/14 12:56] – created naptastic | nndocs:srp [2025/04/15 18:42] (current) – naptastic | ||
---|---|---|---|
Line 1: | Line 1: | ||
====targetcli-fb==== | ====targetcli-fb==== | ||
- | The documentation for SRP is woefully inadequte. | + | * it needs to be turned on in the kernel. drivers -> Infiniband -> SRP |
+ | * turn on both target and initiator please | ||
+ | * in targetcli, /srpt is the relevant path | ||
+ | * targets are formatted like ib.fe80000000000000f4521403002c18b1 | ||
+ | * idk how you get those numbers except to examine an ipoib link, such as: | ||
- | | + | |
- | Missing required parameter wwn | + | |
- | /srpt> ib.fe800000000000005849560e53b70b01/ | + | |
- | WWN not valid as: ib | + | |
- | / | + | |
- | WWN not valid as: ib | + | |
- | From man targetcli(8): | + | Now strip the first 4 bytes off (they change anyway) and remove the :'s |
- | | + | |
- | SRP (SCSI RDMA Protocol) requires that RDMA-capable hardware is present. It uses " | + | |
- | NONE OF THIS TELLS ME HOW TO FORMAT THE STRING. Let' | + | The top 64 bits of the ACL might need to be 0 or they might be the same as the GUID; I don't know the initiator ID actually gets generated and I'm not doing any more digging into it. |
+ | |||
+ | /srpt> ib.fe800000000000005849560e53b70b01/ | ||
+ | Created Node ACL for ib.00000000000000005849560e59150301 | ||
+ | Created mapped LUN 0. | ||
+ | |||
+ | A Linux SRP target is always visible from all InfiniBand partitions. That limits its usefulness to me. | ||
+ | |||
+ | ====Dependencies==== | ||
+ | |||
+ | apt install srptools | ||
+ | |||
+ | Find targets to connect to: | ||
+ | |||
+ | # srp_daemon -o -v -c -p 1 | ||
+ | |||
+ | * -o means "run once" otherwise dmesg on all your hosts will get polluted with SRP login noise. | ||
+ | * -v means "say what you're doing" | ||
+ | * -c means "emit target information in a format we can use later" | ||
+ | * -p 1 means "only scan on HCA port 1" so obviously change this if you are initiating from port 2... | ||
+ | |||
+ | Lazy benchmarking seems good: | ||
+ | |||
+ | [root]@[shark][~]# | ||
+ | 4096+0 records in | ||
+ | 4096+0 records out | ||
+ | 17179869184 bytes (17 GB, 16 GiB) copied, 5.38771 | ||
+ | |||
+ | [root]@[shark][~]# | ||
+ | dd: error writing '/ | ||
+ | 4097+0 records | ||
+ | 4096+0 records out | ||
+ | 17179869184 bytes (17 GB, 16 GiB) copied, 13.7431 s, 1.3 GB/s | ||
+ | |||
+ | ...so I'm using some for swap. We'll see how long it takes me to induce a kernel panic. ;) |
nndocs/srp.1739537778.txt.gz · Last modified: 2025/02/14 12:56 by naptastic