IRC log of #cubox of Tue 29 Mar 2016. All times are in CEST < Back to index

02:12 heap_ hi, i occured strange issue, running kernel 4.4.0 and samba.. when there is a higher load (read) from smb the network stops to work... i mean i can ping eth0 on the server where samba is running but cant ping gateway .. i have to disconnect eth0 and after it everything works again... any idea?
17:37 heap_ hi, i occured strange issue, running kernel 4.4.0 and samba.. when there is a higher load (read) from smb the network stops to work... i mean i can ping eth0 on the server where samba is running but cant ping gateway .. i have to disconnect eth0 and after it everything works again... any idea?
17:49 vpeter heap_: Which distro? Try armbian at least for a test.
17:55 heap_ vpeter: debian
17:55 heap_ vpeter: why armbian?
17:55 vpeter I think it has some extra kernel patches.
17:56 heap_ vpeter: do you think its kernel issue?
17:56 heap_ vpeter: what kernel it has? any link to that armbian distro?
17:56 vpeter I think jnettlet would confirm :)
17:56 vpeter http://www.armbian.com/download/
17:59 heap_ which one should i go for wheezy, jessie, trusty?
18:00 heap_ look like noone plays with 4x kernel ;/
18:01 vpeter wheezy I think. Jessie is stable.
18:02 vpeter Only few people :)
18:11 heap_ hmhm
18:24 jnettlet vpeter, heap, as best as I know all 4.x ports of the kernel have major problems with various aspects
18:26 jnettlet obviously you have your reasons for wanting 4.x however most of our current customers are happily using 3.14.yy. Any work I will put towards 4.4.yy will be regarding upstream enhancements
18:27 heap_ jnettlet: is is possible that there is bug with samba?
18:28 jnettlet heap_, I doubt it. I have had multiple other b2b customers that have reported the same behaviour. As soon I as told them to switch to our Jessie image their problems went away.
18:29 heap_ jnettlet: ok im going to switch to your kernel
18:29 jnettlet heap_, starting with 4.1.yy the upstream NXP maintainer for the ethernet driver has introduced various regressions from build to build. Upstream stability has not been a good thing for us.
18:29 heap_ i really cant stand how "unstable it is" with 4x
18:30 jnettlet heap_, if you still see the problem then feel free to yell at me and I will debug it for you :)
18:30 jnettlet but I have many high bandwidth boards that have been running for months here
18:31 heap_ jnettlet: at the moment my issue is with samba
18:31 heap_ jnettlet: when there is a load (read) then eth stop to work
18:31 heap_ well i can pig eth0 on cubox but not gw, until i reconnect cable manually or if down/up
18:31 jnettlet heap_, yes that is the same problem the customers I worked with reported
18:32 jnettlet I don't think our 3.14.yy kernel has that issue
18:32 heap_ so is it possible to fix in 4x?
18:32 heap_ how many backport received 3.14.60 ?
18:33 jnettlet well I am sure it can be fixed, but the driver has changed a lot in the 4.1.yy and newer kernels
18:37 heap_ hmhm
18:37 heap_ but how
18:39 heap_ jnettlet: can i go with armbian + Kernel 4.4.1 with large hardware support, headers and some firmware included ?
18:39 heap_ vanila
18:39 jnettlet heap_, really your best bet would be to test against a full upstream kernel and then if you run into the same problem report a bug against the upstream kernel
18:39 heap_ but what about vanila Kernel 4.4.1 with large hardware support, headers and some firmware included ?
18:40 jnettlet sure give it a try.
18:40 heap_ hmhm :)
18:40 heap_ ok then
18:40 heap_ are you working on that vanila 4.4.1 also?
18:40 jnettlet although if you run into the bug they will most likely ask you to test against the current mainline
18:40 heap_ or its generic kernel?
18:41 jnettlet for my 4.4.1 testing I generally just test the etnaviv gpu.
18:41 jnettlet nothing that I would consider product quality testing
20:47 vpeter heap_: You have problem with writing on your imx6 system?
21:04 heap_ vpeter: what exactly do you mean?