Skip to content

Commit

Permalink
Use InfiniBandClass from procfs library
Browse files Browse the repository at this point in the history 
Parsing the sysfs files for InfiniBand was added to the procfs library
(see prometheus/procfs#164).

Therefore use `InfiniBandClass` from the procfs library instead of
parsing sysfs itself.

If the port counter return `N/A (no PMA)` no metric will be returned
(instead of returning 0 for this metric.

Signed-off-by: Benjamin Drung <[email protected]>
  • Loading branch information
@bdrung
bdrung committed Aug 28, 2019
1 parent 93c12e0 commit 7a2d79a
Show file tree
Hide file tree
Showing 5 changed files with 149 additions and 250 deletions.
11 changes: 0 additions & 11 deletions collector/fixtures/e2e-64k-page-output.txt
View file
Original file line number Diff line number Diff line change
Expand Up @@ -810,12 +810,10 @@ node_infiniband_legacy_unicast_packets_transmitted_total{device="mlx4_0",port="1
node_infiniband_legacy_unicast_packets_transmitted_total{device="mlx4_0",port="2"} 61239
# HELP node_infiniband_link_downed_total Number of times the link failed to recover from an error state and went down
# TYPE node_infiniband_link_downed_total counter
node_infiniband_link_downed_total{device="i40iw0",port="1"} 0
node_infiniband_link_downed_total{device="mlx4_0",port="1"} 0
node_infiniband_link_downed_total{device="mlx4_0",port="2"} 0
# HELP node_infiniband_link_error_recovery_total Number of times the link successfully recovered from an error state
# TYPE node_infiniband_link_error_recovery_total counter
node_infiniband_link_error_recovery_total{device="i40iw0",port="1"} 0
node_infiniband_link_error_recovery_total{device="mlx4_0",port="1"} 0
node_infiniband_link_error_recovery_total{device="mlx4_0",port="2"} 0
# HELP node_infiniband_multicast_packets_received_total Number of multicast packets received (including errors)
Expand All @@ -828,44 +826,35 @@ node_infiniband_multicast_packets_transmitted_total{device="mlx4_0",port="1"} 16
node_infiniband_multicast_packets_transmitted_total{device="mlx4_0",port="2"} 0
# HELP node_infiniband_port_constraint_errors_received_total Number of packets received on the switch physical port that are discarded
# TYPE node_infiniband_port_constraint_errors_received_total counter
node_infiniband_port_constraint_errors_received_total{device="i40iw0",port="1"} 0
node_infiniband_port_constraint_errors_received_total{device="mlx4_0",port="1"} 0
# HELP node_infiniband_port_constraint_errors_transmitted_total Number of packets not transmitted from the switch physical port
# TYPE node_infiniband_port_constraint_errors_transmitted_total counter
node_infiniband_port_constraint_errors_transmitted_total{device="i40iw0",port="1"} 0
node_infiniband_port_constraint_errors_transmitted_total{device="mlx4_0",port="1"} 0
# HELP node_infiniband_port_data_received_bytes_total Number of data octets received on all links
# TYPE node_infiniband_port_data_received_bytes_total counter
node_infiniband_port_data_received_bytes_total{device="i40iw0",port="1"} 0
node_infiniband_port_data_received_bytes_total{device="mlx4_0",port="1"} 1.8527668e+07
node_infiniband_port_data_received_bytes_total{device="mlx4_0",port="2"} 0
# HELP node_infiniband_port_data_transmitted_bytes_total Number of data octets transmitted on all links
# TYPE node_infiniband_port_data_transmitted_bytes_total counter
node_infiniband_port_data_transmitted_bytes_total{device="i40iw0",port="1"} 0
node_infiniband_port_data_transmitted_bytes_total{device="mlx4_0",port="1"} 1.493376e+07
node_infiniband_port_data_transmitted_bytes_total{device="mlx4_0",port="2"} 0
# HELP node_infiniband_port_discards_received_total Number of inbound packets discarded by the port because the port is down or congested
# TYPE node_infiniband_port_discards_received_total counter
node_infiniband_port_discards_received_total{device="mlx4_0",port="1"} 0
# HELP node_infiniband_port_discards_transmitted_total Number of outbound packets discarded by the port because the port is down or congested
# TYPE node_infiniband_port_discards_transmitted_total counter
node_infiniband_port_discards_transmitted_total{device="i40iw0",port="1"} 0
node_infiniband_port_discards_transmitted_total{device="mlx4_0",port="1"} 5
# HELP node_infiniband_port_errors_received_total Number of packets containing an error that were received on this port
# TYPE node_infiniband_port_errors_received_total counter
node_infiniband_port_errors_received_total{device="i40iw0",port="1"} 0
node_infiniband_port_errors_received_total{device="mlx4_0",port="1"} 0
# HELP node_infiniband_port_packets_received_total Number of packets received on all VLs by this port (including errors)
# TYPE node_infiniband_port_packets_received_total counter
node_infiniband_port_packets_received_total{device="i40iw0",port="1"} 0
node_infiniband_port_packets_received_total{device="mlx4_0",port="1"} 6.825908347e+09
# HELP node_infiniband_port_packets_transmitted_total Number of packets transmitted on all VLs from this port (including errors)
# TYPE node_infiniband_port_packets_transmitted_total counter
node_infiniband_port_packets_transmitted_total{device="i40iw0",port="1"} 0
node_infiniband_port_packets_transmitted_total{device="mlx4_0",port="1"} 6.235865e+06
# HELP node_infiniband_port_transmit_wait_total Number of ticks during which the port had data to transmit but no data was sent during the entire tick
# TYPE node_infiniband_port_transmit_wait_total counter
node_infiniband_port_transmit_wait_total{device="i40iw0",port="1"} 0
node_infiniband_port_transmit_wait_total{device="mlx4_0",port="1"} 4.294967295e+09
# HELP node_infiniband_unicast_packets_received_total Number of unicast packets received (including errors)
# TYPE node_infiniband_unicast_packets_received_total counter
Expand Down
11 changes: 0 additions & 11 deletions collector/fixtures/e2e-output.txt
View file
Original file line number Diff line number Diff line change
Expand Up @@ -810,12 +810,10 @@ node_infiniband_legacy_unicast_packets_transmitted_total{device="mlx4_0",port="1
node_infiniband_legacy_unicast_packets_transmitted_total{device="mlx4_0",port="2"} 61239
# HELP node_infiniband_link_downed_total Number of times the link failed to recover from an error state and went down
# TYPE node_infiniband_link_downed_total counter
node_infiniband_link_downed_total{device="i40iw0",port="1"} 0
node_infiniband_link_downed_total{device="mlx4_0",port="1"} 0
node_infiniband_link_downed_total{device="mlx4_0",port="2"} 0
# HELP node_infiniband_link_error_recovery_total Number of times the link successfully recovered from an error state
# TYPE node_infiniband_link_error_recovery_total counter
node_infiniband_link_error_recovery_total{device="i40iw0",port="1"} 0
node_infiniband_link_error_recovery_total{device="mlx4_0",port="1"} 0
node_infiniband_link_error_recovery_total{device="mlx4_0",port="2"} 0
# HELP node_infiniband_multicast_packets_received_total Number of multicast packets received (including errors)
Expand All @@ -828,44 +826,35 @@ node_infiniband_multicast_packets_transmitted_total{device="mlx4_0",port="1"} 16
node_infiniband_multicast_packets_transmitted_total{device="mlx4_0",port="2"} 0
# HELP node_infiniband_port_constraint_errors_received_total Number of packets received on the switch physical port that are discarded
# TYPE node_infiniband_port_constraint_errors_received_total counter
node_infiniband_port_constraint_errors_received_total{device="i40iw0",port="1"} 0
node_infiniband_port_constraint_errors_received_total{device="mlx4_0",port="1"} 0
# HELP node_infiniband_port_constraint_errors_transmitted_total Number of packets not transmitted from the switch physical port
# TYPE node_infiniband_port_constraint_errors_transmitted_total counter
node_infiniband_port_constraint_errors_transmitted_total{device="i40iw0",port="1"} 0
node_infiniband_port_constraint_errors_transmitted_total{device="mlx4_0",port="1"} 0
# HELP node_infiniband_port_data_received_bytes_total Number of data octets received on all links
# TYPE node_infiniband_port_data_received_bytes_total counter
node_infiniband_port_data_received_bytes_total{device="i40iw0",port="1"} 0
node_infiniband_port_data_received_bytes_total{device="mlx4_0",port="1"} 1.8527668e+07
node_infiniband_port_data_received_bytes_total{device="mlx4_0",port="2"} 0
# HELP node_infiniband_port_data_transmitted_bytes_total Number of data octets transmitted on all links
# TYPE node_infiniband_port_data_transmitted_bytes_total counter
node_infiniband_port_data_transmitted_bytes_total{device="i40iw0",port="1"} 0
node_infiniband_port_data_transmitted_bytes_total{device="mlx4_0",port="1"} 1.493376e+07
node_infiniband_port_data_transmitted_bytes_total{device="mlx4_0",port="2"} 0
# HELP node_infiniband_port_discards_received_total Number of inbound packets discarded by the port because the port is down or congested
# TYPE node_infiniband_port_discards_received_total counter
node_infiniband_port_discards_received_total{device="mlx4_0",port="1"} 0
# HELP node_infiniband_port_discards_transmitted_total Number of outbound packets discarded by the port because the port is down or congested
# TYPE node_infiniband_port_discards_transmitted_total counter
node_infiniband_port_discards_transmitted_total{device="i40iw0",port="1"} 0
node_infiniband_port_discards_transmitted_total{device="mlx4_0",port="1"} 5
# HELP node_infiniband_port_errors_received_total Number of packets containing an error that were received on this port
# TYPE node_infiniband_port_errors_received_total counter
node_infiniband_port_errors_received_total{device="i40iw0",port="1"} 0
node_infiniband_port_errors_received_total{device="mlx4_0",port="1"} 0
# HELP node_infiniband_port_packets_received_total Number of packets received on all VLs by this port (including errors)
# TYPE node_infiniband_port_packets_received_total counter
node_infiniband_port_packets_received_total{device="i40iw0",port="1"} 0
node_infiniband_port_packets_received_total{device="mlx4_0",port="1"} 6.825908347e+09
# HELP node_infiniband_port_packets_transmitted_total Number of packets transmitted on all VLs from this port (including errors)
# TYPE node_infiniband_port_packets_transmitted_total counter
node_infiniband_port_packets_transmitted_total{device="i40iw0",port="1"} 0
node_infiniband_port_packets_transmitted_total{device="mlx4_0",port="1"} 6.235865e+06
# HELP node_infiniband_port_transmit_wait_total Number of ticks during which the port had data to transmit but no data was sent during the entire tick
# TYPE node_infiniband_port_transmit_wait_total counter
node_infiniband_port_transmit_wait_total{device="i40iw0",port="1"} 0
node_infiniband_port_transmit_wait_total{device="mlx4_0",port="1"} 4.294967295e+09
# HELP node_infiniband_unicast_packets_received_total Number of unicast packets received (including errors)
# TYPE node_infiniband_unicast_packets_received_total counter
Expand Down
75 changes: 75 additions & 0 deletions collector/fixtures/sys.ttar
View file
Original file line number Diff line number Diff line change
Expand Up @@ -112,6 +112,21 @@ Mode: 755
Directory: sys/class/infiniband/i40iw0
Mode: 755
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/i40iw0/board_id
Lines: 1
I40IW Board ID
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/i40iw0/fw_ver
Lines: 1
0.2
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/i40iw0/hca_type
Lines: 1
I40IW
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Directory: sys/class/infiniband/i40iw0/ports
Mode: 755
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Expand Down Expand Up @@ -206,9 +221,39 @@ Lines: 1
N/A (no PMA)
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/i40iw0/ports/1/phys_state
Lines: 1
5: LinkUp
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/i40iw0/ports/1/rate
Lines: 1
10 Gb/sec (4X)
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/i40iw0/ports/1/state
Lines: 1
4: ACTIVE
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Directory: sys/class/infiniband/mlx4_0
Mode: 755
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/mlx4_0/board_id
Lines: 1
SM_1141000001000
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/mlx4_0/fw_ver
Lines: 1
2.31.5050
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/mlx4_0/hca_type
Lines: 1
MT4099
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Directory: sys/class/infiniband/mlx4_0/ports
Mode: 755
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Expand Down Expand Up @@ -346,6 +391,21 @@ Lines: 1
0
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/mlx4_0/ports/1/phys_state
Lines: 1
5: LinkUp
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/mlx4_0/ports/1/rate
Lines: 1
40 Gb/sec (4X QDR)
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/mlx4_0/ports/1/state
Lines: 1
4: ACTIVE
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Directory: sys/class/infiniband/mlx4_0/ports/2
Mode: 755
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Expand Down Expand Up @@ -435,6 +495,21 @@ Lines: 1
0
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/mlx4_0/ports/2/phys_state
Lines: 1
5: LinkUp
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/mlx4_0/ports/2/rate
Lines: 1
40 Gb/sec (4X QDR)
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Path: sys/class/infiniband/mlx4_0/ports/2/state
Lines: 1
4: ACTIVE
Mode: 644
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Directory: sys/class/net
Mode: 755
# ttar - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Expand Down
Loading

0 comments on commit 7a2d79a

Please sign in to comment.