Big load on lkml created by -stable patchbombs was Re: [PATCH 5.19 0000/1157] 5.19.2-rc1 review

Big load on lkml created by -stable patchbombs was Re: [PATCH 5.19 0000/1157] 5.19.2-rc1 review

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 2294 bytes --]

Hi!

> > > This is the start of the stable review cycle for the 5.19.2 release.
> > > There are 1157 patches in this series, all will be posted as a response
> > > to this one.  If anyone has any issues with these being applied, please
> > > let me know.
> > 
> > Hi Greg,
> > 
> > Perhaps its time that you just send a single email to LKML pointing where to
> > find the stable releases. These patch bombs are bringing vger down to its
> > knees, and causing delays in people's workflows. This doesn't just affect
> > LKML, but all other vger mailing lists. Probably because LKML has the biggest
> > subscriber base that patch bombs to it can slow everything else down.
> > 
> > I sent 3 patches to the linux-trace-devel list almost 4 hours ago, and they
> > still haven't shown up. I was going to point people to it tonight but it's now
> > going to have to wait till tomorrow.
> 
> Email is async, sometimes it takes longer than others to recieve
> messages.

Well, email is pretty fast most of the month.

> My "patch bombs" get sent out slow to the mail servers, there is work to
> fix up vger and move it over to the LF-managed infrastructure, perhaps
> work with the vger admins to help that effort out?

I'm pretty used to -stable patches going to l-k, so I got used to
current workflow. OTOH ... -stable _is_ quite significant fraction of
total lkml traffic, and I see how people may hate that.

Is not it ultimately for vger admins to decide what kind of load they
consider acceptable?

Would it make sense to somehow batch the messages, or perhaps to
modify patchbombing scripts to send patches "slowly" so that -stable
does not DoS other lkml users?

Actually, if the patch is same between multiple -stable releases
(which is rather common case) sending it just once tagged with "this
goes to 4.9, 4.14, 4.19 and 5.10" would both take less bandwidth and
make review easier. (But I see it may not be that easy).

Best regards,
								Pavel
								
PS: ...who is currently on EDGE connection. LF-managed infrastructure
is not going to help for people who use slow links to access lkml.
-- 
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

Re: Big load on lkml created by -stable patchbombs was Re: [PATCH 5.19 0000/1157] 5.19.2-rc1 review

[Bagas Sanjaya]

[-- Attachment #1: Type: text/plain, Size: 3027 bytes --]

On Tue, Aug 16, 2022 at 01:56:14PM +0200, Pavel Machek wrote:
> Hi!
> 
> > > > This is the start of the stable review cycle for the 5.19.2 release.
> > > > There are 1157 patches in this series, all will be posted as a response
> > > > to this one.  If anyone has any issues with these being applied, please
> > > > let me know.
> > > 
> > > Hi Greg,
> > > 
> > > Perhaps its time that you just send a single email to LKML pointing where to
> > > find the stable releases. These patch bombs are bringing vger down to its
> > > knees, and causing delays in people's workflows. This doesn't just affect
> > > LKML, but all other vger mailing lists. Probably because LKML has the biggest
> > > subscriber base that patch bombs to it can slow everything else down.
> > > 
> > > I sent 3 patches to the linux-trace-devel list almost 4 hours ago, and they
> > > still haven't shown up. I was going to point people to it tonight but it's now
> > > going to have to wait till tomorrow.
> > 
> > Email is async, sometimes it takes longer than others to recieve
> > messages.
> 
> Well, email is pretty fast most of the month.
> 
> > My "patch bombs" get sent out slow to the mail servers, there is work to
> > fix up vger and move it over to the LF-managed infrastructure, perhaps
> > work with the vger admins to help that effort out?
> 
> I'm pretty used to -stable patches going to l-k, so I got used to
> current workflow. OTOH ... -stable _is_ quite significant fraction of
> total lkml traffic, and I see how people may hate that.
> 
> Is not it ultimately for vger admins to decide what kind of load they
> consider acceptable?
> 
> Would it make sense to somehow batch the messages, or perhaps to
> modify patchbombing scripts to send patches "slowly" so that -stable
> does not DoS other lkml users?
> 
> Actually, if the patch is same between multiple -stable releases
> (which is rather common case) sending it just once tagged with "this
> goes to 4.9, 4.14, 4.19 and 5.10" would both take less bandwidth and
> make review easier. (But I see it may not be that easy).
> 

Hi Pavel,

I have to breakdown In-Reply-To chain.

I can't see the message you sent above on lore, so I had to "lei up"
in order to get it. On lore, the patch series thread [1] displayed
is until [1001/1157], and rest of the thread (including yours) is skipped.
Actually, I can see your message from search results [2].

I guess why more than a thousand of patches must be reviewed for this
stable review cycle is because many of them (which have stable list CCed)
are errorneously submitted and merged for 6.0 merge window, not as
stabilization fixes for 5.19. But I also wonder if any of these patches
are not actually qualified for stable (new drivers?).

[1]: https://lore.kernel.org/stable/20220815180439.416659447@linuxfoundation.org/T/#u
[2]: https://lore.kernel.org/stable/20220816115614.GB27428@duo.ucw.cz/

Thanks.

-- 
An old man doll... just what I always wanted! - Clara

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: Big load on lkml created by -stable patchbombs was Re: [PATCH 5.19 0000/1157] 5.19.2-rc1 review

[Steven Rostedt]

On Tue, 16 Aug 2022 13:56:14 +0200
Pavel Machek <pavel@denx.de> wrote:

> I'm pretty used to -stable patches going to l-k, so I got used to
> current workflow. OTOH ... -stable _is_ quite significant fraction of
> total lkml traffic, and I see how people may hate that.

The thing I hate is that this slows down the other mailing lists. It
doesn't just affect LKML. It affects *all* lists. Even the ones that just
get a dozen emails a day. But the emails that do come, are important.

Perhaps another answer is to dedicate the LKML emails to their own servers,
and let the other mailing lists run separately such that the patch bombs to
LKML do not affect any of the other lists.

That would be an acceptable solution for me ;-)

-- Steve