RT-AC68U Out of RAM causing wifi to drop

[Milothicus]

Our office has been running an RT-AC-68U for a couple of years now. we have around 25 devices connected, some wired and some wirelessly, and a couple through the Asus VPN.

we'd had some issues with the wifi dropping recently, so as a first step, i upgraded to the most recent firmware available through the auto-update (3.0.0.4.380_1842).

the problem wasn't solved, and i noted that the ram usage looked very high for very little cpu usage. ram was around 75% while both cpus were below 5%. in researching the issue i read that the newer firmware intentionally uses more ram, but i think the logs indicate that the router is running out of ram. whenever the wifi drops, here's what the log shows repeatedly:

[<c0373504>] (__slab_alloc+0x794/0x814) from [<c0373738>] (__kmalloc_track_caller+0xf8/0x108)
Mar 24 13:07:48 kernel: [<c0373738>] (__kmalloc_track_caller+0xf8/0x108) from [<c0375d28>] (__alloc_skb+0x54/0x104)
Mar 24 13:07:48 kernel: [<c0375d28>] (__alloc_skb+0x54/0x104) from [<c01d8a10>] (osl_alloc_skb+0x18/0x30)
Mar 24 13:07:48 kernel: [<c01d8a10>] (osl_alloc_skb+0x18/0x30) from [<c0385090>] (osl_pktget+0x4c/0x154)
Mar 24 13:07:48 kernel: [<c0385090>] (osl_pktget+0x4c/0x154) from [<c038460c>] (_dma_rxfill+0x174/0x3d4)
Mar 24 13:07:48 kernel: [<c038460c>] (_dma_rxfill+0x174/0x3d4) from [<bf00be30>] (chiprx+0xf0/0x140 [et])
Mar 24 13:07:48 kernel: unwind: Index not found bf00be30
Mar 24 13:07:48 kernel: Mem-info:
Mar 24 13:07:48 kernel: DMA per-cpu:
Mar 24 13:07:48 kernel: CPU    0: hi:   42, btch:   7 usd:  39
Mar 24 13:07:48 kernel: CPU    1: hi:   42, btch:   7 usd:  35
Mar 24 13:07:48 kernel: Normal per-cpu:
Mar 24 13:07:48 kernel: CPU    0: hi:   42, btch:   7 usd:  24
Mar 24 13:07:48 kernel: CPU    1: hi:   42, btch:   7 usd:  19
Mar 24 13:07:48 kernel: active_anon:2849 inactive_anon:2915 isolated_anon:0
Mar 24 13:07:48 kernel:  active_file:917 inactive_file:1026 isolated_file:0
Mar 24 13:07:48 kernel:  unevictable:0 dirty:0 writeback:0 unstable:0
Mar 24 13:07:48 kernel:  free:9497 slab_reclaimable:227 slab_unreclaimable:42804
Mar 24 13:07:48 kernel:  mapped:942 shmem:230 pagetables:177 bounce:0
Mar 24 13:07:48 kernel: DMA free:4144kB min:11020kB low:13772kB high:16528kB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:130048kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:0kB slab_unreclaimable:122624kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? yes
Mar 24 13:07:48 kernel: lowmem_reserve[]: 0 109 109
Mar 24 13:07:48 kernel: Normal free:33844kB min:9456kB low:11820kB high:14184kB active_anon:11396kB inactive_anon:11660kB active_file:3668kB inactive_file:4104kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:111616kB mlocked:0kB dirty:0kB writeback:0kB mapped:3768kB shmem:920kB slab_reclaimable:908kB slab_unreclaimable:48592kB kernel_stack:576kB pagetables:708kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no
Mar 24 13:07:48 kernel: lowmem_reserve[]: 0 0 0
Mar 24 13:07:48 kernel: DMA: 0*4kB 0*8kB 1*16kB 1*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 1*4096kB = 4144kB
Mar 24 13:07:48 kernel: Normal: 21*4kB 22*8kB 11*16kB 2*32kB 5*64kB 4*128kB 3*256kB 0*512kB 1*1024kB 3*2048kB 6*4096kB = 33844kB
Mar 24 13:07:48 kernel: 2173 total pagecache pages
Mar 24 13:07:48 kernel: 0 pages in swap cache
Mar 24 13:07:48 kernel: Swap cache stats: add 0, delete 0, find 0/0
Mar 24 13:07:48 kernel: Free swap  = 0kB
Mar 24 13:07:48 kernel: Total swap = 0kB
Mar 24 13:07:48 kernel: 65536 pages of RAM
Mar 24 13:07:48 kernel: 9824 free pages
Mar 24 13:07:48 kernel: 1607 reserved pages
Mar 24 13:07:48 kernel: 12194 slab pages
Mar 24 13:07:48 kernel: 6352 pages shared
Mar 24 13:07:48 kernel: 0 pages swap cached

This repeats over and over until we reboot the router. then the RAM usage shown on the system status drops from 85% down to about 18%. i did read recently that the memory usage was adjusted to use more ram because it's more efficient, but the logs indicate to me that the router is running out of memory.

is there any other useful info hidden in the log that i'm not recognizing? any ideas of how we might correct this?

 

[RMerlin]

All 380_1xxx firmware releases for the RT-AC68U have problems with wifi (mostly the 2.4 GHz band). Asus has already fixed it since, try switching to the next 380_2xxx releases from them.

 

[Milothicus]

the Asus support website shows that 380_1842 is the mot recent release. is there another place to find the more recent release?

 

[pete y testing]

the Asus support website shows that 380_1842 is the mot recent release. is there another place to find the more recent release?

was going to ask the same thing

 

[RMerlin]

They should have a newer beta release for this model soon (they started releasing these beta builds for other models this week).

Sent from my Nexus 5X using Tapatalk

 

[Milothicus]

I found this in the asus forum:

https://vip.asus.com/forum/view.asp...d_id=11&model=RT-AC68U&page=1&SLanguage=en-us

I've changed 'NAT Acceleration' from 'Disabled' to 'Auto' in the LAN settings, on the 'Switch Control' tab. hopefully this will improve things. I'll report back in a day or so, as the issue came up consistently every day.

 

[RMerlin]

I found this in the asus forum:

https://vip.asus.com/forum/view.asp...d_id=11&model=RT-AC68U&page=1&SLanguage=en-us

I've changed 'NAT Acceleration' from 'Disabled' to 'Auto' in the LAN settings, on the 'Switch Control' tab. hopefully this will improve things. I'll report back in a day or so, as the issue came up consistently every day.

The issue with those broken RT-AC68U builds happen when NAT acceleration is disabled. If you set it to Auto and something causes it to be disabled due to compatibility issues (for instance if you use Traditional QoS), then it will remain disabled, and will still be unstable.

 

[Milothicus]

Thanks for the additional info. The wifi hasn't dropped yet today, so i think things are improved.

hopefully we're not using anything that will result in compatibility issues.