Hi all,
since a few days i´m searching for the reason why the throughput of an SRX240H Cluster is so slow.
additional i debugged and reviewed the whole configuration if there are any problems visible.
Goal: Copy a huge amount of data via VPN to annother Datacenter
Problem:
- Throughput only reaches 90-100 Mbit/s (with Gbit Interface)
(I can see inside monitoring, that the traffic is nearly exact 100 mbit
not many sessions, normal packet count)
- Firewall Internal Traffic is massive delayed (due to high dataplane CPU)
- Checked reth interface VLAN´s
- overall Throughput is roundabout 150 Mbit/s
What i did:
- VPN Tuning (TCP-MSS etc...)
- set low VPN encryption (for testing, no change)
- debugged Flow if there are dropped or anormal packets
- Checked MTU Sizes 1514 (internal)
- Checked Switch Configurations (VLAN, Speed , OK)
- Checked Servers Configuration (Interface Config , Patchlevel, Packettrace, etc. OK)
- Checked Posrtspeed / Duplex etc. OK
- Disabled Logging
- Disabled ALG
- DIsabled UTM
- man other things additional
all i see is:
FPC 0
PIC 0
CPU utilization : 99 %
Memory utilization : 66 %
Current flow session : 393
Current flow session IPv4: 335
Current flow session IPv6: 58
Max flow session : 102400
Total Session Creation Per Second (for last 96 seconds on average): 16
IPv4 Session Creation Per Second (for last 96 seconds on average): 15
IPv6 Session Creation Per Second (for last 96 seconds on average): 1
additional Informaton:
last pid: 63117; load averages: 0.54, 0.27, 0.19 up 8+21:13:04 16:51:04
76 processes: 6 running, 69 sleeping, 1 zombie
CPU states: 77.3% user, 0.0% nice, 1.7% system, 0.0% interrupt, 21.1% idle
Mem: 203M Active, 112M Inact, 557M Wired, 70M Cache, 112M Buf, 29M Free
Swap:
PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU COMMAND
1442 root 139 0 517M 59204K CPU1 1 676.7H 92.48% flowd_octeon_hm
1442 root 139 0 517M 59204K CPU3 3 676.7H 92.48% flowd_octeon_hm
1442 root 139 0 517M 59204K CPU2 2 676.7H 92.48% flowd_octeon_hm
1442 root 80 0 517M 59204K RUN 0 676.7H 8.98% flowd_octeon_hm
63116 root 81 0 8624K 3264K select 0 0:00 0.73% sshd
63117 sshd 8 0 8116K 1784K nanslp 0 0:00 0.73% sshd
1442 root 76 0 517M 59204K select 0 676.7H 0.00% flowd_octeon_hm
1442 root 76 0 517M 59204K select 0 676.7H 0.00% flowd_octeon_hm
1442 root 8 0 517M 59204K nanslp 0 676.7H 0.00% flowd_octeon_hm
1503 root 76 0 28904K 11808K select 0 522:05 0.00% mib2d
1504 root 76 0 21424K 13916K select 0 286:34 0.00% snmpd
1454 root 76 0 12628K 5868K select 0 28:48 0.00% license-check
1495 root 76 0 10720K 4092K select 0 21:38 0.00% nstraced
1474 root 76 0 20516K 9312K select 0 19:41 0.00% l2ald
1477 root 76 0 28380K 14012K select 0 19:30 0.00% kmd
1444 root 76 0 16024K 3764K select 0 13:40 0.00% shm-rtsdbd
1449 root 76 0 13828K 6424K select 0 12:55 0.00% rtlogd
1484 root 76 0 49544K 14160K select 0 12:25 0.00% authd
1432 root 76 0 115M 18240K select 0 9:58 0.00% chassisd
1433 root 76 0 12824K 5172K select 0 9:35 0.00% alarmd
1502 root 76 0 25680K 9552K select 0 9:19 0.00% pfed
1502 root 76 0 25680K 9552K RUN 0 9:19 0.00% pfed
1096 root 76 0 13052K 5208K select 0 8:49 0.00% eventd
1483 root 76 0 50408K 11408K select 0 7:40 0.00% jdhcpd
1498 root 4 0 9632K 4792K kqread 0 7:28 0.00% mcsnoopd
1445 root 76 0 14204K 6996K select 0 6:39 0.00% jsrpd
1429 root 76 0 3304K 1384K select 0 6:13 0.00% bslockd
1480 root 76 0 11728K 5104K select 0 4:30 0.00% dhcpd
1475 root 76 0 14004K 7020K select 0 4:11 0.00% rmopd
1473 root 4 0 52552K 23244K kqread 0 3:52 0.00% rpd
1496 root 76 0 14472K 7092K select 0 3:32 0.00% fwauthd
1451 root 76 0 14184K 4876K select 0 3:12 0.00% wland
is there anybody who has any Idea how to find the core Issue ?
okok, i understand, the reason why the Load is high is caused by the copy job, but why at 100 Mbit /s ?
is it possible to debug the detailed reason for the high CPU load ?
if yes, how ?
Regards
Martin