Eric Wong [Mon, 24 Jun 2019 02:52:16 +0000 (02:52 +0000)]
syscall: get rid of unused EPOLL* constants
EPOLLRDBAND is used for DECnet; and I'm pretty sure I won't be
updating any of our code to work with DECnet.
I've never found use for EPOLLHUP or EPOLLERR, either; so
disable those for now and add comments for things I might
actually use: EPOLLET and EPOLLONESHOT.
Eric Wong [Mon, 24 Jun 2019 02:52:15 +0000 (02:52 +0000)]
ds: get rid of redundant and unnecessary POLL* constants
EPOLL* constants already match their POLL* counterparts and
there's no way Linux can ever diverge or change the values
of those constants. So we'll favor the EPOLL* ones since we
use EPOLLEXCLUSIVE, already.
For weird stuff like kqueue, we'd need to keep maintaining
the mapping, anyways.
Eric Wong [Mon, 24 Jun 2019 02:52:08 +0000 (02:52 +0000)]
ds: split out from ->flush_write and ->write
Get rid of the confusing $need_queue variable and all
the associated documentation for it. Instead, make it
obvious that we're either skipping the write buffer or
flushing the write buffer by splitting the sub in two.
Eric Wong [Mon, 24 Jun 2019 02:52:05 +0000 (02:52 +0000)]
AddTimer: avoid clock_gettime for the '0' case
We rely on immediate timers often, so we can avoid the overhead
of an extra subroutine call to retrieve the monotonic time (and
a sometimes-system call on some platforms).
It never has, AFAIK, but I'm making some changes to this code in
another branch and nearly introduced a bug where it would be
overreading and discarding the pipelined request.
Eric Wong [Sun, 16 Jun 2019 06:11:28 +0000 (06:11 +0000)]
ds: stop distinguishing event read and write callbacks
Having separate read/write callbacks in every class is too
confusing to my easily-confused mind. Instead, give every class
an "event_step" callback which is easier to wrap my head around.
This will make future code to support IO::Socket::SSL-wrapped
sockets easier-to-digest, since SSL_write() can require waiting
on POLLIN events, and SSL_read() can require waiting on POLLOUT
events.
Eric Wong [Sun, 16 Jun 2019 01:04:28 +0000 (01:04 +0000)]
Merge remote-tracking branch 'origin/newspeak' into xcpdb
* origin/newspeak:
comments: replace "partition" with "shard"
t/xcpdb-reshard: use 'shard' term in local variables
xapcmd: favor 'shard' over 'part' in local variables
search: use "shard" for local variable
v2writable: use "epoch" consistently when referring to git repos
adminedit: "part" => "shard" for local variables
v2writable: rename local vars to match Xapian terminology
v2writable: avoid "part" in internal subs and fields
search*: rename {partition} => {shard}
xapcmd: update comments referencing "partitions"
v2: rename SearchIdxPart => SearchIdxShard
inboxwritable: s/partitions/shards/ in local var
tests: change messages to use "shard" instead of partition
v2writable: rename {partitions} field to {shards}
v2writable: count_partitions => count_shards
searchidxpart: start using "shard" in user-visible places
rename reference to git epochs as "partitions"
admin|xapcmd: user-facing messages say "shard"
v2writable: update comments regarding xcpdb --reshard
doc: rename our Xapian "partitions" to "shards"
Eric Wong [Sat, 15 Jun 2019 17:38:42 +0000 (17:38 +0000)]
Merge remote-tracking branch 'origin/ds'
* origin/ds:
ds: stop caring about event flags set by epoll/poll/kqueue
ds: do not distinguish between POLLHUP and POLLERR
ds: remove read method, here, too
nntp: use sysread to append to existing buffer
ds: remove steal_socket method
ds: remove {fd} field
ds: reduce Errno imports and drop ->close reason
ds: cleanup Errno imports and favor constant comparisons
ds: simplify write buffer accounting
Eric Wong [Fri, 14 Jun 2019 10:09:13 +0000 (10:09 +0000)]
t/git-http-backend: explain purpose of test
I found myself tempted to switch to HTTP::Tiny, here, since
it's distributed with Perl since 5.14, unlike Net::HTTP
(which AFAIK was never a part of Perl proper).
But we really want to use Net::HTTP, here, since it's
lower-level and allows us to trigger server-side buffering
by not reading the entity body.
PublicInbox::Inbox objects have minimal dependencies, so
drop code to support old tests which existed before the
PublicInbox::Inbox object came into existence.
Eric Wong [Fri, 14 Jun 2019 16:42:26 +0000 (16:42 +0000)]
t/www_listing: favor HTTP::Tiny over Net::HTTP
More testers are likely to have HTTP::Tiny than Net::HTTP, since
HTTP::Tiny is a dual-life module and distributed with Perl since
Perl 5.14 (2011-05-14), whereas Net::HTTP will likely live in
a separate package forever.
Eric Wong [Fri, 14 Jun 2019 16:25:39 +0000 (16:25 +0000)]
Merge remote-tracking branch 'origin/reshard' into next
* origin/reshard:
xcpdb: support resharding v2 repos
xcpdb: use destination shard as progress prefix
xapcmd: preserve indexlevel based on the destination
v2writable: use a smaller default for Xapian partitions
Eric Wong [Fri, 14 Jun 2019 16:23:13 +0000 (16:23 +0000)]
Merge remote-tracking branch 'origin/manifest' into next
* origin/manifest:
git: ensure ->modified returns an integer
www: support $INBOX/git/$EPOCH.git for v2 cloning
www: wire up /$INBOX/manifest.js.gz, too
wwwlisting: generate grokmirror-compatible manifest.js.gz
wwwlisting: allow hiding entries from manifest
Eric Wong [Thu, 13 Jun 2019 00:29:37 +0000 (00:29 +0000)]
xcpdb: support resharding v2 repos
v2 repos are sometimes created on machines where CPU
parallelization exceeds the capability of the storage devices.
In that case, users may reshard the Xapian DB to any smaller,
positive integer to avoid excessive overhead and contention when
bottlenecked by slow storage.
Resharding can also be used to increase shard count after
hardware upgrades.
Eric Wong [Wed, 12 Jun 2019 00:35:32 +0000 (00:35 +0000)]
v2writable: use a smaller default for Xapian partitions
Apparently 16 CPUs (probably HT) and SATA storage is common
these days. Having excessive Xapian partitions leads to
contention and excessive FD/space use. So set a smaller
default but continue allowing user-specified values to bump
this up.
Eric Wong [Thu, 13 Jun 2019 08:10:02 +0000 (08:10 +0000)]
git: remove cat_file sub callback interface
We weren't using it, and in retrospect, it makes no sense to use
this API cat_file for giant responses which can't read quickly
with minimal context-switching (or sanely fit into memory for
Email::Simple/Email::MIME).
For giant blobs which we don't want slurped in memory, we'll
spawn a short-lived git-cat-file process like we do in ViewVCS.
Otherwise, monopolizing a git-cat-file process for a giant
blob is harmful to other PSGI/NNTP users.
A better interface is coming which will be more suitable for
for batch processing of "small" objects such as commits and
email blobs.
Eric Wong [Fri, 14 Jun 2019 00:27:31 +0000 (00:27 +0000)]
nntp: filter out duplicate Message-IDs for leafnode
It's the unfortunate reality that there are some clients which
reuse Message-IDs (in which we generate + use another) or set
multiple Message-IDs on their own. While the v2 format
addresses that, NNTP clients such as leafnode are not always
prepared to deal with that case.
So, ensure NNTP clients only see a single Message-ID, and
show the others as 'X-Alt-Message-ID'.
Eric Wong [Thu, 13 Jun 2019 20:46:52 +0000 (20:46 +0000)]
nntp: ensure Message-ID is not folded for leafnode
Leafnode cannot handle Message-ID headers which are too long and
require folding via Email::Simple::Header. Since there are
already many of these messages in git with the header already
folded, we need to handle the unfolding when emitting the
message via NNTP.
As far as we know, Leafnode is the only client software
incapable of handling this case.
Eric Wong [Thu, 13 Jun 2019 02:26:30 +0000 (02:26 +0000)]
doc: update dependencies for CentOS 7.x
Digest::SHA is the most notable missing package at runtime
for a minimal system.
Tests don't run at all without Test::Simple (or Test::More).
Plack::Test is also a separate package, too...
Also, the package for IO::Compress::Gzip should be IO::Compress;
as perl-PerlIO-gzip is a different thing entirely which is not
relevant to our needs.
Test::HTTP::Server::Simple doesn't seem required at all for Plack
tests.
ExtUtils::MakeMaker needs to be documented as a install dependency
for people installing this, too; since AFAIK public-inbox is not
yet in any distros.
Eric Wong (Contractor, The Linux Foundation) [Mon, 10 Jun 2019 18:09:27 +0000 (18:09 +0000)]
edit: drop unwanted headers before noop check
mutt will set Content-Length, Lines, and Status headers
unconditionally, so we need to account for that before
doing header comparisons to avoid making expensive changes
when noop edits are made.
Eric Wong [Mon, 10 Jun 2019 03:21:37 +0000 (03:21 +0000)]
ds: stop caring about event flags set by epoll/poll/kqueue
If we got something to write, then write it. Otherwise, try
reading; and continue dealing with errors which normally occur
along the way.
Trying to read requests while we need to buffer in luserspace
is suicidal from a memory management standpoint.
The only adjustment needed for existing callers is EvCleanup;
where we need to ensure we're always calling the dummy
EvCleanup::event_write callback to accomplish nothing.
Eric Wong [Mon, 10 Jun 2019 03:02:11 +0000 (03:02 +0000)]
ds: do not distinguish between POLLHUP and POLLERR
In my experience, both are worthless as any normal read/write
call path will be wanting to check errors and deal with them
appropriately; so we can just call event_read, for now.
Eventually, there'll probably be only one callback for dealing
with all in/out/err/hup events to simplify logic, especially w.r.t
TLS socket negotiation.
Eric Wong [Mon, 10 Jun 2019 02:34:48 +0000 (02:34 +0000)]
nntp: use sysread to append to existing buffer
We already do this in PublicInbox::HTTP, as it's superior to
DS::read in this regard. Initially (when I started writing
NNTP.pm, I wanted to use Danga::Socket's read buffering and
push_back_read (removed in DS) but quickly figured out it wasn't
useful at all for dealing with trickling clients.
Eric Wong [Mon, 10 Jun 2019 02:07:26 +0000 (02:07 +0000)]
ds: remove {fd} field
Storing the file descriptor was redundant as we can quickly call
fileno($self->{sock}) and not have to store an extra hash table
entry. Multiple sources of truth leads to confusion, confusion
leads to bugs.
#!/usr/bin/perl -w
use Benchmark qw(:all);
use Errno qw(EAGAIN EINTR);
my ($r, $w);
pipe($r, $w) or die 'pipe';
require IO::Handle;
$r->blocking(0);
my $buf;
my $n = 30000;
timethese(30, {
hash_hit => sub {
sysread($r, $buf, 1);
for (0..$n) {
next if $!{EAGAIN};
die 'FAIL';
}
}
,
'cmp_eq' => sub {
sysread($r, $buf, 1);
for (0..$n) {
next if $! == EAGAIN;
die 'FAIL';
}
},
hash_miss => sub {
sysread($r, $buf, 1);
for (0..$n) {
die 'FAIL' if $!{EINTR};
}
},
'cmp_ne' => sub {
sysread($r, $buf, 1);
for (0..$n) {
die 'FAIL' if $! == EINTR;
}
},
});
Eric Wong [Mon, 10 Jun 2019 01:33:30 +0000 (01:33 +0000)]
ds: simplify write buffer accounting
Keeping track of write_buf_size was redundant and pointless when
we can simply check the number of elements in the buffer array.
Multiple sources of truth leads to confusion; confusion leads to
bugs.
Finally, rename the prefixes to 'wbuf' to ensure we loudly
(instead of silently) break any external dependencies being
ported over from Danga::Socket, as further changes are pending.
Eric Wong (Contractor, The Linux Foundation) [Sun, 9 Jun 2019 02:51:41 +0000 (02:51 +0000)]
admin: remove warning arg for unconfigured inboxes
We no longer make -index warn on it, no other code uses it;
and working on unconfigured inboxes is totally reasonable
for admins who are setting things up.
Eric Wong (Contractor, The Linux Foundation) [Sun, 9 Jun 2019 02:51:40 +0000 (02:51 +0000)]
v2writable: implement ->replace call
Much of the existing purge code is repurposed to a general
"replace" functionality.
->purge is simpler because it can just drop the information.
Unlike ->purge, ->replace needs to edit existing git commits (in
case of From: and Subject: headers) and reindex the modified
message.
We currently disallow editing of References:, In-Reply-To: and
Message-ID headers because it can cause bad side effects with
our threading (and our lack of rethreading support to deal with
excessive matching from incorrect/invalid References).
Eric Wong (Contractor, The Linux Foundation) [Sun, 9 Jun 2019 02:51:39 +0000 (02:51 +0000)]
import: switch to "replace_oids" interface for purge
Continuing the work by Eric Biederman in commit a118d58a402bd31b
("Import.pm: When purging replace a purged file with a zero length file"),
we can use a generic OID replacement mechanism to implement
purge.
Eric Wong (Contractor, The Linux Foundation) [Sun, 9 Jun 2019 02:51:37 +0000 (02:51 +0000)]
v2writable: consolidate overview and indexing call
It's one ugly sub with lots of parameters, but it's better
than calling a bunch of ugly subs with lots of parameters;
as we'll be needing to call it again when reindexing for
message replacements.