[f-nsp] Max BGP peers BigIron RX4
Pieter Taks
p.taks at nforce.nl
Wed Mar 9 14:15:32 EST 2011
Hi,
Thanks for reading with me and thinking what it could be. Even with all the soft-reconfigurations disabled I still got the same issue. Please see the debug outputs below, perhaps anyone else knows why this is limited to 252 BGP peers on an RX4?
In the sh ip bgp debug variables output it shows Maximum Peer Index Number:253. Is there a way to change this?
sh ip bgp debug variables
safi:0, &bgp:08fe0aac, enabled:1, operational:1, &dbg_mem=08fe1c4c, curr_afi:0
io_process_running:0, io_process_next_peer_number=0
in_long_loops 0, clear_all 0, timer 00000000, count 0
timer_enabled:1, timer_next_peer_number:1, 1s timer 1, short timer 1
scheduler id:5:2, ip:77.247.181.xxx/28, time=92219
bgp_tcb:08fe0b74 (0xf0090001, 0), tick_cnt=12, seconds=6717786
bgp_tcb6:f00a0001 (0x00000000, 0x00000019)
*peer:08fe0bb0, *peer_group:08fe1b5c, RIB_in_root_node:24a35960
Maximum Peer Index Number:253, check_nexthops:0 0
router_id:85.159.239.xx, configured:0, cluster_id:0.0.0.13, configured:1
route_is_router_reflector:0, client_to_client_reflection:1
networks:x08fe401c, aggregate:x08feb0b8
default_metric:4294967294, local_preference:100, keep_alive:60, hold_time:180
originate_default:0, originated:0
distance:20 200 200, fast_external_fallover=0
nexthop recur0, en_def:0, readvertise:1, auto_sum:0, synch:0
always_compare_med:1, compare_med_with_empty_aspath: 0, redistribute_ibgp:150911764, local_network_check_time_count:6
nexthop_cache_hit_count:159342982, nexthop_cache_miss_count:5142
system memory:536870912, total_allocated:104422684, bgp_defined_quota:2147483648
sh ip bgp debug
BGP Debug Information
Pid Size Address Total Used Free NoMem Errors #_pools p_unit
0 8 08fe3d38 43688 31090 12598 0 0 5 2000
1 16 08fe3d64 52426 37260 15166 0 0 5 2000
2 24 08fe3d90 224692 131923 92769 0 0 10 2000
3 32 08fe3dbc 116502 77986 38516 0 0 12 800
4 48 08fe3de8 20163 12144 8019 0 0 7 400
5 64 08fe3e14 3851 981 2870 0 0 5 200
6 96 08fe3e40 2617 397 2220 0 0 6 80
7 128 08fe3e6c 2480 1081 1399 0 0 8 40
8 256 08fe3e98 755 478 277 0 0 6 20
9 512 08fe3ec4 0 0 0 0 0 0 10
10 4096 08fe3ef0 0 0 0 0 0 0 10
11 48 08fe3f1c 362952 349541 13411 0 0 20 4000
12 36 08fe3f48 183498 160937 22561 0 0 8 8000
13 56 08fe3f74 699040 674509 24531 0 0 42 4000
14 84 08fe3fa0 250214 169516 80698 0 0 22 4000
15 96 08fe3fcc 0 0 0 0 0 0 4000
Total Memory Use for Route and Attributes Tables : 104328584
Memory Block Not Available Count : 0
Bad Memory Pool ID Count : 0
TCP buffers : 4096 0 0 0
BGP Tx Parameters : 5 30 3
BGP route update count : 5 (4) last:17h54m45s
event : (5:2) 77.247.181.xxx/28
BGP io semaphore take 54915185, yield 20300844, 14291686
Max timer process: l-5131 s-80740 (463), io: 88163 7431 us
io_rx_yield_time 0x08fe0acc, 1
1 sec timer value: 53364030, 21197655 TB
MP active: 1, standby up 0
Graceful_restart: enable 1, restart time 15, stale-routee 360, purge 600
Restarted 0, fwd 0, restart_up_time_count[0] 0
sh ip bgp debug memory
BGP_CLASS: 198964, BGP_PEER_CLASS: 120071, BGP_CONFIGURATION_CLASS: 3534
BGP_AS_PATH_ENTRY: 84, BGP_IPV6_AS_PATH_ENTRY: 96, BGP_AS_PATH_SEGMENTS: 14
BGP_NLRI_ENTRY: 56, BGP_PATRICIA_KEY_ADDRESS: 16, BGP_PATRICIA_NODE: 48
BGP_RIB_OUT_NLRI_ENTRY: 36, BGP_RIB_OUT_HOLDER: 28, BGP_WITHDRAWN_ROUTE_ENTRY: 34
BGP_NEXTHOP_ADDRESS: 16, BGP_NEXTHOP_ENTRY: 180, BGP_DAMPING_NLRI_ENTRY: 36
BGP_ROUTE_DAMPING_REUSE_LIST: 2052, BGP_ROUTE_DAMPING_BLOCK: 3522, BGP_DAMPENING: 37374
CU_BGP_ROUTE_MAP_SET_T: 234, CU_BGP_ROUTE_MAP_MATCH_T: 1222
BGP_MATCH_TAG: 65, BGP_SET_AS_PATH: 40, BGP_SET_COMMUNITY: 138
BGP_MATCH_UNION: 207, BGP_SET_UNION: 138
ROUTE_FILTER_MEMORY_ENTRY: 220, BGP_MATCH_CRITERIA: 216, BGP_SET_COMMAND: 147
sh ip bgp debug route-table
There are 674476 NLRIs in BGP Route Table, time 323 ms
Best regards,
Pieter Taks
Datum: Fri, 4 Mar 2011 17:03:03 -0600
Onderwerp: Re: [f-nsp] Max BGP peers BigIron RX4
Hi,
That's interesting.
Does the RX series support the 'show ip bgp debug' cmd list? It's hidden up to the point of 'debug'. If you can get that far, take a look around in there, especially 'show ip bgp debug memory' & just the plain 'show ip bgp debug'. Maybe something will stick out.
I want to say the router dynamically allocates memory for certain fields up to a certain dynamic max related to other parts of BGP like # of routes. Do you have soft-reconfig turned on? Maybe try turning that off and see if it frees up any more BGP peers. Brocade software can be a little dodgy at times and not well documented.
Also, I wouldn't be surprised if it was just some sort of bug. We've had stranger things happen when it comes to a bug on the Brocades.
Sincerely.
On Tue, Feb 22, 2011 at 8:23 AM, Pieter Taks <p.taks at nforce.nl<mailto:p.taks at nforce.nl>> wrote:
Hi all,
While configuring a new peer on an BigIron RX4 we got the following error:
Error! BGP4 cannot allocate memory for peer 253
A PDF claims the following (http://www.terabitsystems.com/foundry-docs/Foundry-BigIron-RX-Series-Datasheet.pdf):
BGPv4: Scalable to 4 million routes, 500 peers and 14,000 attributes with MR2 management module
Is there any reason why it wouldn’t want to configure more peering sessions?
- I do not see any system-max setting for this
- As well as I do not see any full memory yet
Hope someone knows how to resolve this or what is causing this error.
sh memory
====================================================================
BigIron RX active MP slot 33:
Total SDRAM: 2147483648 bytes
Available Memory: 1646874624 bytes
Free Physical Pages: 401374 pages
Malloc statistics: total 500462813
os_malloc count: 24059060, fail: 3; os_free count: 24046183, fail 0, diff: 12877
====================================================================
BigIron RX LP SL 1:
Total SDRAM: 536870912 bytes
Available Memory: 53039104 bytes
====================================================================
BigIron RX LP SL 3:
Total SDRAM: 536870912 bytes
Available Memory: 72671232 bytes
====================================================================
BigIron RX LP SL 4:
Total SDRAM: 536870912 bytes
Available Memory: 72654848 bytes
Any help is appreciated, thanks!
Pieter Taks
_______________________________________________
foundry-nsp mailing list
foundry-nsp at puck.nether.net<mailto:foundry-nsp at puck.nether.net>
http://puck.nether.net/mailman/listinfo/foundry-nsp
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/foundry-nsp/attachments/20110309/6175d094/attachment.html>
More information about the foundry-nsp
mailing list