I think James' concern is on the HTTP TPC only, not regular HTTP request. I would suggest that James submit an enhancement request (not to xrootd-l but to xrootd github) so that we can keep track of this.
regards,
--
Wei Yang | [log in to unmask] | 650-926-3338(O)
-----Original Message-----
From: <[log in to unmask]> on behalf of Andrew Hanushevsky <[log in to unmask]>
Date: Monday, January 10, 2022 at 11:26 AM
To: Alessandra Forti <[log in to unmask]>
Cc: Petr Vokac <[log in to unmask]>, "Yang, Wei" <[log in to unmask]>, James William Walder <[log in to unmask]>, xrootd-l <[log in to unmask]>
Subject: Re: Max connections via davs tpc ?
Well, only one site wanted a limit and we came up with a bypass for them,
I believe. That was a long time ago. So, it's not that we can't, it's that
we haven't had any requests to do so up until now. As far as I can
understand, the discussion here is only for TPC. For xroot TPC, there is a
limit. I suppose we can figure out how to do that for the http side,
though it's a bit more complicated there because of the implementation.
Andy
On Mon, 10 Jan 2022, Alessandra Forti wrote:
> Hi Petr,
>
> interesting DPM can limit the number of clients but naked xrootd can't what
> is the difference since DPM uses xrootd underneath? We should concentrate on
> the storage types that will remain beyond 2024.
>
> cheers
>
> alessandra
>
> On 10/01/2022 13:26, Petr Vokac wrote:
>> But e.g. DPM limit is by default 1200 per each disknode, we observed
>> problems with dCache sites with their too low default HTTP movers limit, so
>> I'm basically happy if there is no limit on XRootD HTTP side (till somebody
>> try to actively kill our services, but robust protection of our
>> infrastructure would be completely different task).
>>
>> Petr
>>
>> On 1/10/22 12:54, Alessandra Forti wrote:
>>> Hi Wei,
>>>
>>> is there a reason why it is not possible to set the number of connections?
>>> If not shouldn't this be implemented? Standard apache servers have a
>>> MaxClients directive AFAIK.
>>>
>>> The problem with FTS limitation is that we have several independent FTS in
>>> our community and putting a limit in all of them can be a problem.
>>>
>>> cheers
>>>
>>> alessandra
>>>
>>> On 07/01/2022 19:11, Yang, Wei wrote:
>>>> I don't think HTTP TPC respect that limit. One thing you can ask is for
>>>> FTS service to put a limit on the number of concurrent HTTP TPC
>>>> transfers. The FTS limits apply upon protocol://host but not port.
>>>>
>>>> regards,
>>>> --
>>>> Wei Yang | [log in to unmask] | 650-926-3338(O)
>>>>
>>>> -----Original Message-----
>>>> From: <[log in to unmask]> on behalf of James William Walder
>>>> <[log in to unmask]>
>>>> Date: Friday, January 7, 2022 at 5:39 AM
>>>> To: xrootd-l <[log in to unmask]>
>>>> Subject: Max connections via davs tpc ?
>>>>
>>>> Hi,
>>>> In the ofs.tpc directive the xfr parameter should control the max
>>>> number of TPC connections.
>>>> Does tpc via http respect that parameter ? If not, is there a way to
>>>> set a limit?
>>>> Thanks in advance,
>>>> James
>>>> ########################################################################
>>>> Use REPLY-ALL to reply to list
>>>> To unsubscribe from the XROOTD-L list, click the following
>>>> link:
>>>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
>>>>
>>>>
>>>> ########################################################################
>>>> Use REPLY-ALL to reply to list
>>>>
>>>> To unsubscribe from the XROOTD-L list, click the following link:
>>>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
>>>
>>> ########################################################################
>>> Use REPLY-ALL to reply to list
>>>
>>> To unsubscribe from the XROOTD-L list, click the following link:
>>> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
>>
>
> ########################################################################
> Use REPLY-ALL to reply to list
>
> To unsubscribe from the XROOTD-L list, click the following link:
> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
>
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the XROOTD-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=XROOTD-L&A=1
|