Date | Commit message (Collapse) |
|
This reuses some of the configuration from -watch, but remains
independent since some configurations will use -watch for some
inboxes and -mda for others.
The default remains "spamc" for -mda users so nothing changes
without explicit configuration.
Per-inbox configurations may also be supported in the future.
|
|
It's a convenient wrapper nowadays, so get rid of some legacy
code and minimize differences from the -watch code.
|
|
I've hit some case where probabilistic searches don't work when
using dfpre:/dfpost:/dfblob: search prefixes because stemming in
the query parser interferes.
In any case, our indexing code indexes longer/unabbreviated blob
names down to its 7 character abbreviation, so there should be
no need to do wildcard searches on git blob names.
|
|
For v1 repos, we don't need to write any metadata to Xapian
and changing from 'basic' to 'medium' or 'full' will work.
For v2, the metadata for indexing is stored in msgmap (because
the Xapian databases are partitioned for parallelism), so a
reindex is required.
|
|
Use ||= '' to ensure that if the From or Sender header is not present
the code sees an empty string and instead of undefined.
I had some email messages with a From field without an @ (because the
sender was local) and without a Sender which were causing errors when
imported. I think this was bad enough that the email messages were
failing to be imported.
Signed-off-by: Eric Biederamn <ebiederm@xmission.com>
|
|
Xapian documents and respect XAPIAN_FLUSH_THRESHOLD to define
the interval in documents to flush, so don't override it with
our own BATCH_BYTES. This is helpful for initial indexing for
those on slower storage but enough RAM.
It is unnecessary for -watch and frequent incremental indexing;
and it increases transaction times if -watch is playing "catch-up"
if it was stopped for a while.
The original BATCH_BYTES was tuned for a machine with little
memory as the default XAPIAN_FLUSH_THRESHOLD of 10000 documents
was causing swap storms. Using document counts also proved an
innaccurate estimator of RAM usage compared to the actual bytes
processed.
|
|
This adds a new inbox configuration option 'indexlevel' that can take
the values 'full', 'medium', and 'basic'.
When set to 'full' everything is indexed including the positions
of all terms.
When set to 'medium' everything except the positions of terms is
indexed.
When set to 'basic' terms and positions are not indexed. Just the
Overview database for NNTP is created. Which is still quite good and
allows searching for messages by Message-ID. But there are no indexes to support
searching inside the email messages themselves.
Update the reindex tests to exercise the full medium and basic code paths
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
|
|
Create a new method add_xapian that holds all of the code to create
Xapian indexes. The creation of this method simpliy involved
idenitifying the relevant code and moving it from add_message.
A call is added to add_xapian from add_message to keep everything
working as it currently does. The new call is made conditional upon
index levels of 'full' and 'medium'. The index levels that index
positions and terms the two things public-inbox uses Xapian to index.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
|
|
About half the size of the Xapian search index turns out to be search
positions. The search positions are only used in a very narrow set of
queries. Make the search positions optional so people don't need to
pay the cost of queries they will never make.
This also makes public-inbox more approachable for light hacking as
generating all of the indexes is time consuming.
The way this is done is to add a method to SearchIdx called index_text
that wraps the call of the term generator method index_text. The new
index_text method takes care of calling both index_text and
increase_termpos (the two functions that are responsible for position
data).
Then index_users, index_diff_inc, index_old_diff_fn, index_diff,
index_body are made proper methods that calls the new index_text.
Callers of the new index_text are slightly simplified as they don't
need to call increase_termpos as well.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
|
|
Decrement regen_down when visiting messages that appear in %D that we
know will later be deleted. This ensures consistent message numbers are
generated no matter which commit number is on top. Allowing deletes to
propagage separately from the messages they delete without causing
problems.
The v2 trees already do this and when the indexes are deleted and
rebuilt they maintain they commit numbers.
Add a v1 version of the v2reindex test to verify that reindexing is
working properly on v1 as well as v2.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
|
|
The normal behavior is to prevent the deleted messages from
being indexed in the first place. However, when fetching
incrementally via git; public-inbox-index needs to account for
deleted files which were created outside of the most recent
fetch/reindexing window.
Reported-by: Eric W. Biederman <ebiederm@xmission.com>
|
|
Recently I ran git --git-dir=lkml/git/1.git fsck
and it reported:
> warning in commit 299dbd50b6995c6debe2275f0df984ce697fb4cc: nulInCommit: NULL byte inthe commit object body
Which I found quite scary. Nulls in the wrong place have a bad tendency
to make programs misbehave.
It turns out someone had placed "=?iso-8859-1?q?=00?=" at the end of
their subject line. Which is the mime encoding for NULL. Email::Mime
had correctly decoded the header, and then public-inbox had simply
copied the contents of the header into the subject line of the git
commit.
To prevent that from causing problems replace nulls in such subject
lines with spaces.
Signed-off-by: Eric Biederman <ebiederm@xmission.com>
|
|
Recently I had trouble cloning lkml/git/0.git because
git fsck on receive was failing. The output of git fsck was:
> Checking object directories: 100% (256/256), done.
> warning in commit 59173dc1fe67b113ace4ce83e7f522414b3e0404: badTimezone: invalid author/committer line - bad time zone
> warning in commit ff22aaff22eb4479e49e93f697e385f76db51c55: badTimezone: invalid author/committer line - bad time zone
> warning in commit 609b744909693f5f00aff5ed9928beeeee9ded2e: badTimezone: invalid author/committer line - bad time zone
> warning in commit 084572141db8e0d879428afb278bd338f2dbb053: badTimezone: invalid author/committer line - bad time zone
> warning in commit 789d204de27cd12c6da693d903390a241a1a4bca: badTimezone: invalid author/committer line - bad time zone
> warning in commit 0d9a65948b0c957007ca387cd56b690f9bab9c08: badTimezone: invalid author/committer line - bad time zone
> warning in commit f7468c42b4196ee6323afb373ab9323971c38d69: badTimezone: invalid author/committer line - bad time zone
> warning in commit 85e0cd6dd527cd55ad0440f14384529b83818228: badTimezone: invalid author/committer line - bad time zone
> warning in commit f31e19a2e772c9ed00728ef142af9c550ea5de6a: badTimezone: invalid author/committer line - bad time zone
> warning in commit 56eb7384443ef84e17e29504a304a071b189ae67: badTimezone: invalid author/committer line - bad time zone
> warning in commit e4470030471e6810414b9de5e3b52e16f2245d12: badTimezone: invalid author/committer line - bad time zone
> warning in commit f913b48caa097c3b2cb3f491707944f88d52d89f: badTimezone: invalid author/committer line - bad time zone
> warning in commit 4390f26923d572c6dab6cce8282c7cad5520d785: badTimezone: invalid author/committer line - bad time zone
> warning in commit 0f66db71a06bd7d651a0cd80877d8043b70fda20: badTimezone: invalid author/committer line - bad time zone
> warning in commit d71472c40b36dcdf0396afc9778f6137eea45887: badTimezone: invalid author/committer line - bad time zone
> warning in commit e8d3b19a91a2d86b6a91bd19dc811e851398b519: badTimezone: invalid author/committer line - bad time zone
> warning in commit afd9fc0cc87e56ed7736d633e17d0ef77817b3cc: badTimezone: invalid author/committer line - bad time zone
> warning in commit 811b3217708358cf1b75fba4602a64a426fce0f5: badTimezone: invalid author/committer line - bad time zone
> warning in commit e7a751a597c6f5e4770c61bdee6220d55a37cba9: badTimezone: invalid author/committer line - bad time zone
> warning in commit 3e32ad6192fe093e03e6b9346c3a90b16d9905c0: badTimezone: invalid author/committer line - bad time zone
> warning in commit 5e66b47528e79d3bbb769e137f036a1fa99cccf9: badTimezone: invalid author/committer line - bad time zone
> warning in commit d90d67d94ca47142670dff13fcb81ab7afab07bb: badTimezone: invalid author/committer line - bad time zone
> Checking objects: 100% (1711464/1711464), done.
> Checking connectivity: 1711464, done.
Upon examination with git show --pretty=raw all of the problem commits
had a time zone that was not 4 digits long. This time zone had been
passed straight from the Date line in the email into the author line
of the commit.
Looking into that I discovered that str2time takes into account the
time zone, and was actually able to process these weird time zones.
So get the normalized time zone with strptime and convert it from
seconds from gmt to hours and minutes from gmt.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
|
|
For v2 repositories with multiple epochs, we must not forget
about earlier epochs in clones. Ensure we update the alternates
file with all known epochs up to the current one.
Reported-by: Eric W. Biederman <ebiederm@xmission.com>
https://public-inbox.org/meta/871scj2vzi.fsf@xmission.com/
|
|
In PSGI, PATH_INFO contains URI-decoded paths which cause
problems when Message-IDs contain ambiguous characters for used
for routing. Instead, extract the undecoded path from
REQUEST_URI and use that.
Reported-by: Leah Neukirchen <leah@vuxu.org>
https://public-inbox.org/meta/8736xsb5s5.fsf@vuxu.org/
|
|
The query planner in sqlite3 < 3.8 is not very clever, so when it sees
num mentioned in the query filter, it decides not to use the fast idx_ts
index and goes for the much slower autoindex. CentOS-7 still has
sqlite-3.7, so loading the http landing page of a very large archive
(LKML) was taking over 18 seconds, as oppposed to milliseconds on a
system with sqlite-3.8 and above:
$ time sqlite3 -line over.sqlite3 'SELECT ts,ds,ddd FROM over \
WHERE num > 0 ORDER BY ts DESC LIMIT 1000;' > /dev/null
real 0m19.610s
user 0m17.805s
sys 0m1.805s
$ sqlite3 -line over.sqlite3 'EXPLAIN QUERY PLAN SELECT ts,ds,ddd \
FROM over WHERE num > 0 ORDER BY ts DESC LIMIT 1000;'
selectid = 0
order = 0
from = 0
detail = SEARCH TABLE over USING INDEX sqlite_autoindex_over_1 (num>?) (~250000 rows)
However, if we slightly tweak the query per SQlite recommendations [1]
by adding + to the num filter, we force it to use the correct index
and see much faster performance:
$ time sqlite3 -line over.sqlite3 'SELECT ts,ds,ddd FROM over \
WHERE +num > 0 ORDER BY ts DESC LIMIT 1000;' > /dev/null
real 0m0.007s
user 0m0.005s
sys 0m0.002s
$ sqlite3 -line over.sqlite3 'EXPLAIN QUERY PLAN SELECT ts,ds,ddd \
FROM over WHERE +num > 0 ORDER BY ts DESC LIMIT 1000;'
selectid = 0
order = 0
from = 0
detail = SCAN TABLE over USING INDEX idx_ts (~1464303 rows)
This appears to be the only place where this is needed in order to avoid
running into this issue.
As far as I can tell, this change has no impact on systems running newer
sqlite3 (>= 3.8).
.. [1] https://sqlite.org/optoverview.html#disqualifying_where_clause_terms_using_unary_
Signed-off-by: Konstantin Ryabitsev <konstantin@linuxfoundation.org>
|
|
This is consistent with git itself and the previous behavior
was a result of misunderstanding of how git interprets this.
And adjust tests slightly to match the new behavior.
Reported-by: Konstantin Ryabitsev <konstantin@linuxfoundation.org>
<38873789-ab42-65a1-20c9-12c30b171f4f@linuxfoundation.org>
|
|
Xapian v1.2.21..v1.2.24 (inclusive) use OFD locks but failed to
set the close-on-exec flag on those locks. So we must continue
to work around those old versions by ensuring Xapian file
descriptors aren't held any longer than necessary when in
long-running git processes.
Reported-by: Konstantin Ryabitsev <konstantin@linuxfoundation.org>
|
|
Xapian will replace files upon committing, so non-parallel
V2Writable users need to have umask preserved this way.
|
|
Improve the display by finding any parent when we see out-of-order
References. This prevents us from having two roots in the test
case like Mail::Thread does.
|
|
In retrospect, the loop prevention done by our indexer is not
always sufficient since it can have an improperly sorted
or incomplete References headers.
This bug was triggered multiple bracketed Message-IDs in an
In-Reply-To: header (not References) where the Message-IDs were
in non-chronological order when somebody tried to reply to
different leafs of a thread with a single message.
So we must check for descendents before blindly trying to
use the last one.
Fixes: c6a8fdf71e2c336f ("thread: last Reference always wins")
|
|
All callers in expect to iterate through results. This
was causing unfairness when fetching large ranges via XHDR
as rtin does :<
Fixes: b8c41362f2a5c8fc "nntp: simplify the long_response API"
|
|
Previous search queries already set sort order on the Enquire
object, altering the ordering of results and was causing
messages to be redundantly downloaded via POST /$INBOX/?q=$QUERY&x=m
So stop caching the Search::Xapian::Enquire object since it
wasn't providing any measurable performance improvement.
|
|
It's ugly and all of our other parameters are omitted
when values are not the default.
|
|
We no longer need to parse and dedupe References:
ourselves, PublicInbox::MID::references does it for us.
|
|
It is common to have large amounts of addresses Cc:-ed in large
mailing lists like LKML. Make them more readable by wrapping
after addresses. Unfortunately, line breaks inserted by the
MUA get lost when using the public Email::MIME API.
Subject and body lines remain unwrapped, as it's the author's
fault to have such long lines :P
|
|
The old loop did not help with code clarity with the various
conditional statements. It also hid a bug where we forgot to
(optionally) obfuscate email addresses in Subject: lines if
search was enabled.
|
|
"LIKE" in SQLite (and other SQL implementations I've seen) is
expensive with nearly 3 million messages in the archives.
This caused some partial Message-ID lookups to take over 600ms
on my workstation (~300ms on a faster Xeon). Cut that to below
under 30ms on average on my workstation by relying exclusively
on Xapian for partial Message-ID lookups as we have in the past.
Unlike in the past when we tried using Xapian to match partial
Message-IDs; we now optimize our indexing of Message-IDs to
break apart "words" in Message-IDs for searching, yielding
(hopefully) "good enough" accuracy for folks who get long URLs
broken across lines when copy+pasting.
We'll also drop the (in retrospect) pointless stripping of
"/[tTf]" suffixes for the partial match, since anybody who
hits that codepath would be hitting an invalid message ID.
Finally, limit wildcard expansion to prevent easy DoS vectors
on short terms.
And blame Pine and alpine for generating Message-IDs with
low-entropy prefixes :P
|
|
I was using this to trace the path of brian's message.
Fixes: 017fed7bc4d33ac4
("searchidx: regenerate and avoid article number gaps on full index")
|
|
Relaxing this lock during a v1 --reindex is important to keep
messages showing up in -watch process in a timely manner.
Looks like I deleted an extra line when doing the following
for v2:
s/xdb->commit_transaction/self->commit_txn_lazy/
Fixes: 35ff6bb106909b1c ("replace Xapian skeleton with SQLite overview DB")
|
|
For Subject/To/Cc/From headers, we squeeze them to a space (' ').
For Message-IDs (including References/In-Reply-To), '\t', '\n', '\r'
are deleted since some MUAs might screw them up:
https://public-inbox.org/git/656C30A1EFC89F6B2082D9B6@localhost/raw
|
|
This should reduce idle cat-file instances
|
|
I suppose it's a bug or inconsistency that altid is write-only
and their deletions do not get reflected. But for now, we
do not set it when training spam so there's no window where
an invalid NNTP article number shows up.
This should solve the problem where there's massive gaps
in messages solved by spam training for ruby groups:
https://public-inbox.org/meta/20180307093754.GA27748@dcvr/
|
|
We can't have files with permissions inconsistent with what's
in git objects.
|
|
* origin/master:
nntp: allow and ignore empty commands
mbox: do not barf on queries which return no results
nntp: fix NEWNEWS command
searchview: fix non-numeric comparison
Allow specification of the number of search results to return
githttpbackend: avoid infinite loop on generic PSGI servers
http: fix modification of read-only value
extmsg: use news.gmane.org for Message-ID lookups
extmsg: rework partial MID matching to favor current inbox
Update the installation instructions with Fedora package names
nntp: do not drain rbuf if there is a command pending
nntp: improve fairness during XOVER and similar commands
searchidx: do not modify Xapian DB while iterating
Don't use LIMIT in UPDATE statements
|
|
This increases indexing time by around 10% but roughly
halves memory usage of an -index process.
We will probably make this tunable in the future for people
with bigger/smaller machines.
|
|
Somebody hitting "\n" into telnet shouldn't hold a client up
indefinitely and prevent shutdown.
|
|
We do not want phrase searches to cross between independent
fields (filenames/Message-ID vs bodies)
|
|
We generally do not want git to waste time finding abbreviations
and we do not want the possibility of them becoming ambiguous
over time, either.
|
|
Searching across different inboxes is expensive without
SQLite (or Xapian) installed, so avoid doing expensive tree
lookups in git. Since SQLite is required for Xapian
support anyways, we won't need to check Xapian, either.
Sites without SQLite installed will simply 404 if somebody
requests a message which isn't in the current inbox.
|
|
Some messages to git@vger went missing from Msgmap from old bugs
and became inaccessible via NNTP. Forcing NNTP article numbers
when the overview DB came about made the problem more visible when
reindexing old (v1) repositories as all removed spam messages
took up AUTOINCREMENT numbers again before they were removed.
Having large gaps in NNTP article numbers is not good since it
throws off NNTP clients. This does NOT prevent NNTP clients from
seeing some messages twice, but is better than having them
miss several messages entirely.
We also avoid depending on --reverse in git-log, as
git requires storing an entire commit list in memory for
--reverse, so it's cheaper to store only deleted blobs in the %D
hash since they do not live long.
|
|
In case people were running old buggy versions from 2016...
(and -convert should probably clean those up, eventually)
|
|
First off, decode text portions of messages since some archived
mail I got was converted from quoted-printable or base-64 to
8bit by the original recipient. Attempting to merge them with
my own archives (which had no conversion done) led to
unnecessary duplicates showing up.
Then, normalize CRLF line endings in text portions to LF.
In the headers, we relax the content_id hashing to ignore quotes
and lower-case domain names in To, Cc, and From headers since
some mail processors will alter them.
Finally, I've discovered Email::MIME->new($mime->as_string)
does not always round-trip reliably, so we calculate the
content_id twice on user-supplied messages.
|
|
While hunting duplicates, I noticed a leading '-' in some
Message-IDs as a result of RFC4648 encoding. While '-' seems
allowed by RFC5322 and URL-friendly (RFC4648), they are uncommon
and make using Message-IDs as arguments for command-line tools
more difficult. So prefix them with a datestamp to at least
give readers some sense of the age. And shorten the "localhost"
hostname to "z" to save space.
|
|
I'm not sure how useful this view is, but it exists for now.
|
|
git fast-import and the main V2Writable process combined takes
about one CPU, so avoid having too many Xapian partitions which
cause unnecessary I/O contention.
|
|
Otherwise articles show up again...
|
|
Gigantic feeds probably make some clients unhappy,
clamp it to what it was in the past.
Fixes: b9534449ecce2c59 ("view: avoid offset during pagination")
|
|
This significantly improves the performance of the NNTP GROUP
command with 2.7 million messages from over 250ms to 700us.
SQLite is weird about this, but at least there's a way to
optimize it.
|
|
Since we only query the SQLite over DB for OVER/XOVER; do not
need to waste space storing fields To/Cc/:bytes/:lines or the
XNUM term. We only use From/Subject/References/Message-ID/:blob
in various places of the PSGI code.
For reindexing, we will take advantage of docid stability
in "xapian-compact --no-renumber" to ensure duplicates do not
show up in search results. Since the PSGI interface is the
only consumer of Xapian at the moment, it has no need to
search based on NNTP article number.
|