iscsi
- https://linux.vbird.org/linux_server/centos6/0460iscsi.php
openiscsi 和 libiscsi
- kernel 实现 openiscsi ，配合用户态 iscsid 使用。
- libiscsi 是用户态的，所以每次构建 qemu 会有问题
open-iscsi 是这个: https://github.com/open-iscsi/open-iscsi
基本原理
- https://wiki.nix-pro.com/view/ISCSI
targetcli 和 openiscsi 的基本使用
https://manpages.ubuntu.com/manpages/focal/man8/targetcli.8.html
目前找到的最好操作方法:
- https://www.cnblogs.com/xiangsikai/p/10876534.html
- 服务器端的配置，无需步骤 8 ，同样可以正常运行。
在 openeuler 中测试结果如下:
server
sudo yum -y install targetcli
执行 targetcli
执行 ls
o- / .................................................... [...]
o- backstores ......................................... [...]
| o- block ............................. [Storage Objects: 0]
| o- fileio ............................ [Storage Objects: 0]
| o- pscsi ............................. [Storage Objects: 0]
| o- ramdisk ........................... [Storage Objects: 0]
o- iscsi ....................................... [Targets: 0]
o- loopback .................................... [Targets: 0]
o- srpt ........................................ [Targets: 0]
o- vhost ....................................... [Targets: 0]
o- xen-pvscsi .................................. [Targets: 0]
显然，backstores 就是当前的后端， iscsi 等可以提供的前端。
创建后端
# 进入指定目录下
/> cd /backstores/block
# 创建共享设备，命名为 “disk0”
/backstores/block> create disk0 /dev/nvme1n1
o- / ....................................................................... [...]
o- backstores ............................................................ [...]
| o- block ................................................ [Storage Objects: 1]
| | o- disk0 .................... [/dev/nvme1n1 (1.5TiB) write-thru deactivated]
| | o- alua ................................................. [ALUA Groups: 1]
| | o- default_tg_pt_gp ..................... [ALUA state: Active/optimized]
| o- fileio ............................................... [Storage Objects: 0]
| o- pscsi ................................................ [Storage Objects: 0]
| o- ramdisk .............................................. [Storage Objects: 0]
o- iscsi .......................................................... [Targets: 0]
o- loopback ....................................................... [Targets: 0]
o- srpt ........................................................... [Targets: 0]
o- vhost .......................................................... [Targets: 0]
o- xen-pvscsi ..................................................... [Targets: 0]
当然最简单的是构建一个文件来作为后端:
$ sudo targetcli
/> backstores/fileio create test /tmp/test.img 100m
/> iscsi/ create iqn.2006-04.com.example:test-target
/> cd iscsi/iqn.2006-04.com.example:test-target/tpg1/
tpg1/> luns/ create /backstores/fileio/test
tpg1/> set attribute generate_node_acls=1
tpg1/> exit
创建前端
cd /iscsi
/iscsi> create
o- / ..................................................................................................................... [...]
o- backstores .......................................................................................................... [...]
| o- block .............................................................................................. [Storage Objects: 1]
| | o- disk0 .................................................................... [/dev/nvme1n1 (1.5TiB) write-thru activated]
| | o- alua ............................................................................................... [ALUA Groups: 1]
| | o- default_tg_pt_gp ................................................................... [ALUA state: Active/optimized]
| o- fileio ............................................................................................. [Storage Objects: 0]
| o- pscsi .............................................................................................. [Storage Objects: 0]
| o- ramdisk ............................................................................................ [Storage Objects: 0]
o- iscsi ........................................................................................................ [Targets: 1]
| o- iqn.2003-01.org.linux-iscsi.localhost.aarch64:sn.980b9a90bf36 ................................................. [TPGs: 1]
| o- tpg1 ........................................................................................... [no-gen-acls, no-auth]
| o- acls ...................................................................................................... [ACLs: 0]
| o- luns ...................................................................................................... [LUNs: 1]
| | o- lun0 .............................................................. [block/disk0 (/dev/nvme1n1) (default_tg_pt_gp)]
| o- portals ................................................................................................ [Portals: 1]
| o- 0.0.0.0:3260 ................................................................................................. [OK]
o- loopback ..................................................................................................... [Targets: 0]
o- srpt ......................................................................................................... [Targets: 0]
o- vhost ........................................................................................................ [Targets: 0]
o- xen-pvscsi ................................................................................................... [Targets: 0]
关联 backstores 和前端:
# 进入指定目录
/iscsi> cd /iscsi/iqn.2003-01.org.linux-iscsi.localhost.x8664:sn.037746bdcf2d/tpg1/luns
# 创建共享设备到 target
/iscsi/iqn.20...f2d/tpg1/luns> create /backstores/block/disk0
关闭认证
/> cd /iscsi/iqn.2003-01.org.linux-iscsi.localhost.aarch64:sn.980b9a90bf36/tpg1
/iscsi/.../tpg1> set attribute authentication=0
/iscsi/.../tpg1> set attribute generate_node_acls=1
/iscsi/.../tpg1> set attribute demo_mode_write_protect=0
client 操作
iscsiadm 还需要服务 iscsid 的服务
sudo yum install open-iscsi
sudo systemctl enable iscsid
sudo systemctl start iscsid
🍎 sudo iscsiadm -m discovery -t st -p 10.0.0.2
10.0.0.2:3260,1 iqn.2003-01.org.linux-iscsi.localhost.aarch64:sn.980b9a90bf36
🍎 sudo iscsiadm -m node -T iqn.2003-01.org.linux-iscsi.localhost.aarch64:sn.980b9a90bf36 -p 10.0.0.2:3260 --login
然后可以得到:
lrwxrwxrwx - root 5 Mar 23:42 sdb -> ../../devices/platform/host1/session1/target1:0:0/1:0:0:0/block/sdb
iscsiadm -m discovery -t st -p 127.0.0.1
iscsiadm -m node -T iqn.2003-01.org.linux-iscsi.localhost.x8664:sn.037746bdcf2d -p 127.0.0.1 --login
常见命令
- iscsiadm -m session : 枚举所有的 session 的内容
内核 iscsi 代码简单分析
client 侧
主要的代码在: drivers/scsi/iscsi_tcp.c drivers/scsi/libiscsi.c drivers/scsi/libiscsi_tcp.c
相当于 iscsi 也是一个 hba 卡了:
static const struct scsi_host_template iscsi_sw_tcp_sht = {
.module = THIS_MODULE,
.name = "iSCSI Initiator over TCP/IP",
.queuecommand = iscsi_queuecommand,
.change_queue_depth = scsi_change_queue_depth,
.can_queue = ISCSI_TOTAL_CMDS_MAX,
.sg_tablesize = 4096,
.max_sectors = 0xFFFF,
iscsi_target_transport 只是 iscsit_transport 的一种实现，也就是 TCP/IP 来实现的。
例如 iscsit_get_rx_pdu，可以找到如下三个:
drivers/infiniband/ulp/isert/ib_isert.c
2603: .iscsit_get_rx_pdu = isert_get_rx_pdu,
drivers/target/iscsi/cxgbit/cxgbit_main.c
684: .iscsit_get_rx_pdu = cxgbit_get_rx_pdu,
drivers/target/iscsi/iscsi_target.c
679: .iscsit_get_rx_pdu = iscsit_get_rx_pdu,
- do_softirq
- __do_softirq
- net_rx_action
- napi_poll
- __napi_poll
- process_backlog
- __netif_receive_skb_one_core
- ip_local_deliver_finish
- ip_protocol_deliver_rcu
- tcp_v4_rcv
- tcp_v4_do_rcv
- tcp_rcv_established
- tcp_data_queue
- iscsi_sw_tcp_data_ready
- iscsi_sw_tcp_recv_data
- tcp_read_sock
- iscsi_sw_tcp_recv
- iscsi_tcp_recv_skb
- iscsi_tcp_process_data_in
- iscsi_complete_pdu
- __iscsi_complete_pdu
- iscsi_data_in_rsp
- __iscsi_put_task
- iscsi_free_task
- scsi_done
- scsi_done_internal
- blk_mq_complete_request
- blk_mq_complete_request_remote
- blk_mq_complete_send_ipi
- smp_call_function_single_async
- generic_exec_single
- __smp_call_single_queue
- __send_ipi_mask
- __smp_call_single_queue
- generic_exec_single
- smp_call_function_single_async
- blk_mq_complete_send_ipi
- blk_mq_complete_request_remote
- blk_mq_complete_request
- scsi_done_internal
- scsi_done
- iscsi_free_task
- __iscsi_put_task
- iscsi_data_in_rsp
- __iscsi_complete_pdu
- iscsi_complete_pdu
- iscsi_tcp_process_data_in
- iscsi_tcp_recv_skb
- iscsi_sw_tcp_recv
- tcp_read_sock
- iscsi_sw_tcp_recv_data
- iscsi_sw_tcp_data_ready
- tcp_data_queue
- tcp_rcv_established
- tcp_v4_do_rcv
- tcp_v4_rcv
- ip_protocol_deliver_rcu
- ip_local_deliver_finish
- __netif_receive_skb_one_core
- process_backlog
- __napi_poll
- napi_poll
- net_rx_action
- __do_softirq
@[
iscsi_queuecommand+0
blk_mq_dispatch_rq_list+260
__blk_mq_sched_dispatch_requests+1172
blk_mq_sched_dispatch_requests+56
blk_mq_run_hw_queue+388
blk_mq_dispatch_list+308
blk_mq_flush_plug_list+88
__blk_flush_plug+248
blk_finish_plug+64
read_pages+312
page_cache_ra_unbounded+316
force_page_cache_ra+176
page_cache_sync_ra+72
filemap_get_pages+248
filemap_read+268
blkdev_read_iter+140
aio_read.constprop.0+240
io_submit_one.constprop.0+408
__arm64_sys_io_submit+220
invoke_syscall.constprop.0+88
do_el0_svc+72
el0_svc+92
el0t_64_sync_handler+268
el0t_64_sync+408
]: 5657
@[
virtnet_poll_tx+0
net_rx_action+364
handle_softirqs+304
__do_softirq+28
____do_softirq+24
call_on_irq_stack+48
do_softirq_own_stack+36
do_softirq+152
__local_bh_enable_ip+336
__dev_queue_xmit+676
ip_finish_output2+1764
__ip_finish_output+176
ip_finish_output+60
ip_output+120
__ip_queue_xmit+420
ip_queue_xmit+28
__tcp_transmit_skb+1184
tcp_write_xmit+1148
__tcp_push_pending_frames+68
tcp_push+144
tcp_sendmsg_locked+800
tcp_sendmsg+64
inet_sendmsg+76
sock_sendmsg+136
iscsi_sw_tcp_xmit_segment+164
iscsi_sw_tcp_pdu_xmit+132
iscsi_tcp_task_xmit+88
iscsi_xmit_task+96
iscsi_xmitworker+216
process_one_work+540
worker_thread+452
kthread+340
ret_from_fork+16
]: 19
server 侧
主要的代码在: drivers/target/
当时这个 target 应该是在 virtio-scsi 的盘上
- common_interrupt
- __common_interrupt
- handle_irq
- generic_handle_irq_desc
- handle_edge_irq
- handle_irq_event
- handle_irq_event_percpu
- __handle_irq_event_percpu
- vring_interrupt
- vring_interrupt
- virtscsi_req_done
- virtscsi_vq_done
- scsi_io_completion
- scsi_end_request
- blk_update_request
- req_bio_endio
- iblock_complete_cmd
- target_complete_cmd
- target_complete_cmd_with_sense
- target_complete_cmd
- iblock_complete_cmd
- req_bio_endio
- blk_update_request
- scsi_end_request
- scsi_io_completion
- virtscsi_vq_done
- virtscsi_req_done
- vring_interrupt
- vring_interrupt
- __handle_irq_event_percpu
- handle_irq_event_percpu
- handle_irq_event
- handle_edge_irq
- generic_handle_irq_desc
- handle_irq
- __common_interrupt
- ret_from_fork
- kthread
- worker_thread
- process_one_work
- target_complete_ok_work
- iscsit_add_cmd_to_response_queue
- target_complete_ok_work
- process_one_work
- worker_thread
- kthread
target 侧发送数据:
- ret_from_fork
- kthread
- iscsi_target_tx_thread
- iscsit_handle_response_queue
- iscsit_response_queue
- iscsit_send_datain
- iscsit_xmit_pdu
- iscsit_xmit_datain_pdu
- iscsit_fe_sendpage_sg
- tx_data
- sock_sendmsg
- sock_sendmsg_nosec
- tcp_sendmsg
- tcp_sendmsg_locked
- tcp_push
- tcp_sendmsg_locked
- tcp_sendmsg
- sock_sendmsg_nosec
- sock_sendmsg
- tx_data
- iscsit_fe_sendpage_sg
- iscsit_xmit_datain_pdu
- iscsit_xmit_pdu
- iscsit_send_datain
- iscsit_response_queue
- iscsit_handle_response_queue
- iscsi_target_tx_thread
- kthread
用 scsi_debug 做 iscsi 的后端，有问题?
[ 1576.866364] Unable to recover from DataOut timeout while in ERL=0, closing iSCSI connection for I_T Nexus iqn.2003-01.org.linux-iscsi.localhost.x8664:sn.a6b106aa84c2:xsk,i,0x00023d000004,iqn.2003-01.org.linux-iscsi.localhost.x8664:sn.a6b106aa84c2,t,0x01
[ 1584.148092] connection4:0: ping timeout of 5 secs expired, recv timeout 5, last rx 4296241093, last ping 4296246144, now 4296251392
[ 1584.148816] connection4:0: detected conn error (1022)
[ 1601.556022] iSCSI Login timeout on Network Portal 0.0.0.0:3260
[ 1617.428294] Did not receive response to NOPIN on CID: 0, failing connection for I_T Nexus iqn.2003-01.org.linux-iscsi.localhost.x8664:sn.a6b106aa84c2:xsk,i,0x00023d000004,iqn.2003-01.org.linux-iscsi.localhost.x8664:sn.a6b106aa84c2,t,0x01
[ 1704.980737] session4: session recovery timed out after 120 secs
[ 1704.984660] sd 6:0:0:0: rejecting I/O to offline device
[ 1704.984883] I/O error, dev sdf, sector 401408 op 0x1:(WRITE) flags 0x2004000 phys_seg 128 prio class 2
[ 1704.985221] I/O error, dev sdf, sector 399360 op 0x1:(WRITE) flags 0x2004000 phys_seg 128 prio class 2
[ 1704.985543] I/O error, dev sdf, sector 398336 op 0x1:(WRITE) flags 0x2004000 phys_seg 128 prio class 2
[ 1704.985874] I/O error, dev sdf, sector 397312 op 0x1:(WRITE) flags 0x2004000 phys_seg 128 prio class 2
[ 1704.986199] I/O error, dev sdf, sector 285696 op 0x1:(WRITE) flags 0x2000000 phys_seg 128 prio class 2
[ 1704.986528] I/O error, dev sdf, sector 284672 op 0x1:(WRITE) flags 0x2004000 phys_seg 128 prio class 2
[ 1704.986534] I/O error, dev sdf, sector 530432 op 0x1:(WRITE) flags 0x2004000 phys_seg 128 prio class 2
[ 1704.986579] I/O error, dev sdf, sector 531456 op 0x1:(WRITE) flags 0x2000000 phys_seg 128 prio class 2
[ 1704.986582] I/O error, dev sdf, sector 532480 op 0x1:(WRITE) flags 0x2004000 phys_seg 128 prio class 2
[ 1704.986584] I/O error, dev sdf, sector 533504 op 0x1:(WRITE) flags 0x2004000 phys_seg 128 prio class 2
[ 1704.986590] Write-error on swap-device (8:80:535552)
[ 1704.986598] Write-error on swap-device (8:80:662528)
[ 1704.986599] Write-error on swap-device (8:80:666624)
[ 1704.986605] Write-error on swap-device (8:80:670720)
[ 1704.986606] Write-error on swap-device (8:80:674816)
[ 1704.986608] Write-error on swap-device (8:80:793600)
[ 1704.986610] Write-error on swap-device (8:80:797696)
[ 1704.986611] Write-error on swap-device (8:80:801792)
[ 1704.986616] Write-error on swap-device (8:80:1059840)
[ 1704.986616] Write-error on swap-device (8:80:805888)
[ 1704.987260] iscsi_trx (1489) used greatest stack depth: 12736 bytes left
[ 1704.990926] iSCSI Login negotiation failed.
[ 1704.991308] iSCSI Login negotiation failed.
[ 1704.991478] iSCSI Login negotiation failed.
[ 1704.991641] iSCSI Login negotiation failed.
[ 1704.991806] iSCSI Login negotiation failed.
[ 1704.991966] iSCSI Login negotiation failed.
[ 1704.992127] iSCSI Login negotiation failed.
[ 1704.992486] stress-ng (1609) used greatest stack depth: 12352 bytes left
看来是是网络的问题吧，只要这样配置就会有问题吗？
是不可以通过网络吗?
[ 1926.637286] sd 5:0:0:0: [sde] tag#148 CDB: Read(10) 28 00 00 19 93 d8 00 00 08 00
[ 1926.637547] I/O error, dev sde, sector 1676248 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2
[ 1926.637843] Read-error on swap-device (8:64:1676248)
[ 1926.638025] sd 5:0:0:0: [sde] tag#149 FAILED Result: hostbyte=DID_TIME_OUT driverbyte=DRIVER_OK cmd_age=124s
[ 1926.638379] sd 5:0:0:0: [sde] tag#149 CDB: Read(10) 28 00 00 02 35 c0 00 00 08 00
[ 1926.638639] I/O error, dev sde, sector 144832 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2
[ 1926.638931] Read-error on swap-device (8:64:144832)
[ 1926.639102] sd 5:0:0:0: [sde] tag#150 FAILED Result: hostbyte=DID_TIME_OUT driverbyte=DRIVER_OK cmd_age=121s
[ 1926.639452] sd 5:0:0:0: [sde] tag#150 CDB: Read(10) 28 00 00 04 77 a0 00 00 08 00
[ 1926.639711] I/O error, dev sde, sector 292768 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2
[ 1926.640002] Read-error on swap-device (8:64:292768)
[ 1926.640180] sd 5:0:0:0: rejecting I/O to offline device
[ 1926.640317] I/O error, dev sde, sector 1084688 op 0x1:(WRITE) flags 0x2000000 phys_seg 1 prio class 2
[ 1926.640366] I/O error, dev sde, sector 39184 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2
[ 1926.640896] Write-error on swap-device (8:64:1084688)
[ 1926.641791] Read-error on swap-device (8:64:39184)
[ 1926.642676] I/O error, dev sde, sector 1113808 op 0x1:(WRITE) flags 0x2000000 phys_seg 1 prio class 2
[ 1926.643217] Write-error on swap-device (8:64:1113808)
[ 1926.643395] I/O error, dev sde, sector 1486552 op 0x1:(WRITE) flags 0x2000000 phys_seg 1 prio class 2
[ 1926.643890] Write-error on swap-device (8:64:1486552)
[ 1926.644115] I/O error, dev sde, sector 397880 op 0x1:(WRITE) flags 0x2000000 phys_seg 1 prio class 2
[ 1926.644430] Write-error on swap-device (8:64:397880)
[ 1926.644617] I/O error, dev sde, sector 589592 op 0x1:(WRITE) flags 0x2000000 phys_seg 1 prio class 2
[ 1926.645294] Write-error on swap-device (8:64:589592)
[ 1926.645527] I/O error, dev sde, sector 601712 op 0x1:(WRITE) flags 0x2000000 phys_seg 1 prio class 2
[ 1926.645953] Write-error on swap-device (8:64:601712)
[ 1926.646127] Write-error on swap-device (8:64:649976)
[ 1926.679777] Read-error on swap-device (8:64:1200712)
[ 1926.681659] Read-error on swap-device (8:64:1200736)
[ 1926.681812] htop (1453) used greatest stack depth: 11088 bytes left
[ 1926.685040] Read-error on swap-device (8:64:1356296)
[ 1926.685273] Read-error on swap-device (8:64:1078848)
[ 1926.685728] Read-error on swap-device (8:64:1078448)
[ 1926.685907] Read-error on swap-device (8:64:1080016)
对于 fio 的盘分析一下
测试的 client 和 server 在一起的:
- 14.34% do_syscall_64 ▒
- 13.80% __x64_sys_io_uring_enter ▒
- 9.06% io_submit_sqes ▒
- 8.98% io_issue_sqe ▒
- 8.82% io_read ▒
- __io_read ▒
- 8.79% blkdev_read_iter ▒
- 8.71% blkdev_direct_IO ▒
- 7.62% submit_bio_noacct_nocheck ▒
- 7.48% __submit_bio ▒
- 6.55% __blk_flush_plug ▒
- 6.55% blk_mq_flush_plug_list ▒
- 6.04% blk_mq_run_hw_queue ▒
- 5.98% blk_mq_sched_dispatch_requests ▒
- 5.98% __blk_mq_sched_dispatch_requests ▒
- 4.82% blk_mq_dispatch_rq_list ▒
- 4.81% scsi_queue_rq ▒
- 4.09% iscsi_queuecommand ▒
3.47% queue_work_on ▒
0.88% blk_mq_submit_bio
- 25.87% 0.00% iscsi_ttx [kernel.kallsyms] [k] ret_from_fork_asm ▒
ret_from_fork_asm ▒
ret_from_fork ▒
kthread ▒
- iscsi_target_tx_thread ▒
- 24.06% iscsit_response_queue ▒
- 23.74% iscsit_xmit_pdu ▒
- 23.69% iscsit_fe_sendpage_sg ▒
- 18.28% tcp_sendmsg ▒
- 16.76% tcp_sendmsg_locked ▒
- 16.26% __tcp_push_pending_frames ▒
- tcp_write_xmit ▒
- 15.76% __tcp_transmit_skb ▒
- 15.54% __ip_queue_xmit ▒
- 15.42% ip_finish_output2 ▒
- 15.37% __dev_queue_xmit ▒
- 15.06% __local_bh_enable_ip ▒
- do_softirq ▒
- handle_softirqs ▒
- 10.14% blk_complete_reqs ▒
- 9.61% scsi_io_completion ▒
- scsi_end_request ▒
- 8.20% blk_update_request ▒
- 7.71% blkdev_bio_end_io_async ▒
- 7.11% __io_req_task_work_add ▒
- 7.03% try_to_wake_up ▒
7.00% _raw_spin_unlock_irqrestore
- 4.54% net_rx_action ▒
- 4.47% __napi_poll ▒
- process_backlog ▒
- 4.20% __netif_receive_skb_one_core ▒
- 4.06% ip_local_deliver ▒
- 3.81% ip_protocol_deliver_rcu ▒
- 3.77% tcp_v4_rcv ▒
- 3.28% tcp_v4_do_rcv ▒
- tcp_rcv_established ▒
- 3.00% tcp_data_queue ▒
- 2.66% iscsi_sw_tcp_data_ready ▒
- 2.63% tcp_read_sock ▒
- 1.44% iscsi_sw_tcp_recv ▒
- 1.41% iscsi_tcp_recv_skb ▒
- 0.99% iscsi_tcp_process_data_in ▒
- 0.96% iscsi_complete_pdu ▒
- 0.72% __iscsi_complete_pdu ▒
0.56% iscsi_free_task ▒
0.61% __tcp_transmit_skb ▒
- 1.42% release_sock ▒
- 1.37% __release_sock ▒
- 1.26% tcp_v4_do_rcv ▒
- 1.25% tcp_rcv_established ▒
0.86% tcp_ack ▒
- 5.32% tx_data ▒
- 5.23% sock_sendmsg ▒
- 5.17% tcp_sendmsg ▒
- 4.98% tcp_sendmsg_locked ▒
- 4.58% __tcp_push_pending_frames ▒
- tcp_write_xmit ▒
- 4.14% __tcp_transmit_skb
- 3.87% __ip_queue_xmit ▒
- 3.75% ip_finish_output2 ▒
- 3.66% __dev_queue_xmit ▒
- 3.29% __local_bh_enable_ip ▒
- do_softirq ▒
- handle_softirqs ▒
- net_rx_action ▒
- 2.84% __napi_poll ▒
- process_backlog ▒
- 2.62% __netif_receive_skb_one_core ▒
- 2.52% ip_local_deliver ▒
- 2.39% ip_protocol_deliver_rcu ▒
- 2.36% tcp_v4_rcv ▒
- 2.11% tcp_v4_do_rcv ▒
- 2.09% tcp_rcv_established ▒
- 1.18% tcp_data_queue ▒
- 0.89% iscsi_sw_tcp_data_ready ▒
- 0.81% tcp_read_sock ▒
- 0.55% iscsi_sw_tcp_recv ▒
0.52% iscsi_tcp_recv_skb ▒
- 0.99% schedule ▒
- __schedule ▒
finish_task_switch.isra.0
其他
https://github.com/Masorubka1/iscsi-client-rs
不知道为什么性能很差:
trash: (g=0): rw=randread, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=libaio, iodepth=64
fio-3.39
Starting 1 process
^Cbs: 1 (f=1): [r(1)][0.6%][r=8336KiB/s][r=2084 IOPS][eta 16m:34s]
注意，后端是一个 nvme 的裸盘:
本站所有文章转发 CSDN 将按侵权追究法律责任，其它情况随意。