On 09/09/17 20:23, Jamal Hadi Salim wrote:
> On 17-09-09 12:24 PM, Roopa Prabhu wrote:
>> On Fri, Sep 8, 2017 at 2:52 PM, Roman Mashak <m...@mojatatu.com> wrote:
>>> Signed-off-by: Roman Mashak <m...@mojatatu.com>
>>> ---
>>> bridge/link.c | 16 +++++++++++++---
>>> 1 file changed, 13 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/bridge/link.c b/bridge/link.c
>>> index 60200f1..9e4206f 100644
>>> --- a/bridge/link.c
>>> +++ b/bridge/link.c
>>> @@ -461,9 +461,19 @@ static int brlink_show(int argc, char **argv)
>>> }
>>> }
>>>
>>> - if (rtnl_wilddump_request(&rth, PF_BRIDGE, RTM_GETLINK) < 0) {
>>> - perror("Cannon send dump request");
>>> - exit(1);
>>> + if (show_details) {
>>> + if (rtnl_wilddump_req_filter(&rth, PF_BRIDGE, RTM_GETLINK,
>>> + (compress_vlans ?
>>> +
>>> RTEXT_FILTER_BRVLAN_COMPRESSED :
>>> + RTEXT_FILTER_BRVLAN)) < 0) {
>>> + perror("Cannon send dump request");
>>> + exit(1);
>>> + }
>>
>> vlan information is already available with `bridge vlan show`. any
>> specific reason why you want it in
>> the link dump output ?
>>
>>
>> The problem is this might just make the link dump larger and also add
>> too much clutter into the regular link dump output. iproute2 detailed
>> dump is already a bit hard to interpret. And without compression by
>> default, vlan info can just take over the link dump output. It will be
>> hard to look for other link attributes after that :). We deploy with
>> thousands of vlans and without compression even bridge vlan default
>> output is already hard to interpret.
>>
>
> Agree we should be turning on this stuff by default. i.e default stays
> compressed; otherwise it a huge dump.
I think this should be dumped with the getlink request only on some additional
flag. The getlink does not include these by default.
>
> Having said that there is a lot of mess with this stuff.
> The bridge link events _already send this IFLA_AF_SPCE info_
> so not much choice there but to print it.
Right, on NEWLINK per port notification you'll get the compressed vlan info.
> At minimal we need that part because unfortunately there is no
> vlanfilter event in existence which will send us summaries of just
> vlans added to a port i.e both use XXXLINK.
But let's either add a new flag or use -compressvlans to print it when
monitoring/showing
link otherwise people who are monitoring only the port flags will start getting
lists
with vlans. Even compressed these can still be quite long and confusing,
especially
when monitoring.
> In general, the XXXLINK interface with these master devices (bridge,
> bond etc) continues to get messy. Recently started seeing events with
> devices claiming to be of KIND_slave etc on the bridge and bond devices;
> yet at the same time my wireless card events are also showing up on the
> bridge link even though it is not enslaved there..
>
> cheers,
> jamal
>
>> Can you please paste a sample default detailed output with your patch
>> with a few hundred vlans ?
>>
>
