Regular process crashed in BNG ASR9K

[somnathsafi2344]

Need to understand where i can check the below:

1.Tracking the process which are crashed

2.Reason of crashed process

3.Same ASR9K is being used for the IPMPLS routing and BNG,so whether i can differentiate the process by some command or by referring to any IDs

4.Is iEDGE process crashed can led to wrong accounting in the Radius counter

[smilstea]

1. You can track via syslog which processes have crashed, specifically the 'DUMPER' syslogs. You can also look in harddisk:/dumper for crash files.

2. The signal number is printed, this will indicate whether it was a memory leak or segmentation fault etc, but in general you need to contact TAC to get a decode of the crash.

3. Its really knowledge of the platform to know what will impact traffic or not, for example fib_mgr crashing will impact traffic, but depends if its on just one LC or the RP, etc on the whole impact. There is no exhaustive list of what processes have an impact.

4. This sounds oddly specific, I would recommend recording a TAC case so they can investigate the crash, but in general iedge crashes will have an impact on BNG.

Sam

[mivens]

For 1) the command "show context" will also give you the last crash for each location.

[somnathsafi2344]

Is it possible to manually trigger a scenario which may led to crash of any process in lab environment specially like to check the iEDGE process stability or its threshold to counter the increasing memory demand of the process.

[mivens · Accepted solution]

There's the "process crash" or "process restart" command if you want to check the impact of a certain process in a lab environment. There's the "show memory compare" command if you want to compare memory snapshosts to look for memory leaks over time and "show memory-snapshots process". On newer versions of IOS-XR, PAM will automatically detect increases in memory utilisation of a process and collect diagnostic information that can be used by the TAC. There's also a watchdog process that will automatically restart a process if necessary.

[somnathsafi2344]

Thanks for the reply,

Kindly also provide your valuable comment on the below mentioned area,as i am not getting such answer on the release notes of 6.5.3

RSP has a memory of 32 GB and the same is being allocated to every process running in the BNG, whether the process pick the dynamic memory or the certain threshold or range is pre-define at OS level for each process.

Thanks in advance

[smilstea]

You are running 32-bit XR so the max that can be allocated is 4GB.

We need to look at the dynamic and dynamic limit, dynamic is how much memory is currently used and dyn-limit is how much memory the process can use total.

RP/0/RSP0/CPU0:ASR9010-A#show process fib_mgr
Tue Oct 1 16:10:42.617 UTC
Job Id: 233
PID: 352436
Process name: fib_mgr

<>

RP/0/RSP0/CPU0:ASR9010-A#show process memory detail 233
Tue Oct 1 16:07:48.252 UTC
JID Text Data Stack Dynamic Dyn-Limit Shm-Tot Phy-Tot Process
------ ---------- ---------- ---------- ---------- ---------- ---------- ---------- -------
233 3M 1M 656K 4M 1664M 1228M 5M fib_mgr
Total text: 964 pages
data: 256 pages
stack: 164 pages
malloced: 1050 pages

RP/0/RSP0/CPU0:ASR9010-A#show process memory
Tue Oct 1 16:08:28.006 UTC
JID Text(KB) Data(KB) Stack(KB) Dynamic(KB) Process
------ ---------- ---------- ---------- ----------- -------
61 122880 61440 147456 155975680 eth_server
365 131072 77824 143360 68702208 nvgen_server
1191 200704 86016 36864 64290816 policy_repository
315 1196032 655360 217088 45481984 l2fib_mgr
1137 30670848 6987776 360448 35147776 netconf
1064 6950912 4743168 229376 32632832 te_control
1130 491520 786432 45056 31350784 lspv_server
373 172032 147456 188416 30822400 parser_server
1066 1290240 11108352 544768 28119040 bgp
1186 462848 663552 466944 27099136 ipv4_rib
453 4096 8192 131072 26046464 sysdb_shared_sc
1188 462848 155648 155648 24264704 ipv6_rib
456 4096 8192 73728 23035904 sysdb_svr_local
89 126976 45056 102400 20844544 redfs_svr
1223 3764224 1675264 217088 20389888 l2vpn_mgr
1062 1757184 704512 106496 18952192 rsvp
252 704512 733184 159744 18190336 gsp
1226 40960 24576 20480 16257024 schema_server
1201 3219456 1650688 294912 14753792 pim
1202 3280896 1675264 217088 14557184 pim6
1183 4096 8192 45056 13344768 statsd_manager_g

RP/0/RSP0/CPU0:ASR9010-A#show process memory | i fib_mgr
Tue Oct 1 16:11:35.430 UTC
315 1196032 655360 217088 45481984 l2fib_mgr
233 3948544 1048576 671744 4300800 fib_mgr

Sam

[somnathsafi2344]

Thanks for the reply,

Kindly also mentioned

1.whether i can check the max and min memory which a particular process uses.

2.Whether a minimum threshold memory is allocated to each process or it is allocated dynamically

3.How the RSP Card memory relation with the process memory

Thanks in advance