New inboxes: public-inbox-init -V2

If you're starting a new inbox (and not mirroring an existing one), the -V2 requires DBD::SQLite, but is orders of magnitude more scalable than the original "-V1" format.

Optional Inline::C use

Our optional use of Inline::C speeds up subprocess spawning from large daemon processes.

To enable Inline::C, either set the "PERL_INLINE_DIRECTORY" environment variable to point to a writable directory, or create "~/.cache/public-inbox/inline-c" for any user(s) running public-inbox processes.

If libgit2 development files are installed and Inline::C is enabled (described above), per-inbox "git cat-file --batch" processes are replaced with a single perl(1) process running "PublicInbox::Gcf2::loop" in read-only daemons. libgit2 use will be available in public-inbox 1.7.0+

More (optional) Inline::C use will be introduced in the future to lower memory use and improve scalability.

Note: Inline::C is required for lei(1), but not public-inbox-*

Performance on rotational hard disk drives

Random I/O performance is poor on rotational HDDs. Xapian indexing performance degrades significantly as DBs grow larger than available RAM. Attempts to parallelize random I/O on HDDs leads to pathological slowdowns as inboxes grow.

While "-V2" introduced Xapian shards as a parallelization mechanism for SSDs; enabling "publicInbox.indexSequentialShard" repurposes sharding as mechanism to reduce the kernel page cache footprint when indexing on HDDs.

Initializing a mirror with a high "--jobs" count to create more shards (in "-V2" inboxes) will keep each shard smaller and reduce its kernel page cache footprint. Keep in mind excessive sharding imposes a performance penalty for read-only queries.

Users with large amounts of RAM are advised to set a large value for "publicinbox.indexBatchSize" as documented in public-inbox-index(1).

"dm-crypt" users on Linux 4.0+ are advised to try the "--perf-same_cpu_crypt" "--perf-submit_from_crypt_cpus" switches of cryptsetup(8) to reduce I/O contention from kernel workqueue threads.

Btrfs (and possibly other copy-on-write filesystems)

btrfs(5) performance degrades from fragmentation when using large databases and random writes. The Xapian + SQLite indices used by public-inbox are no exception to that.

public-inbox 1.6.0+ disables copy-on-write (CoW) on Xapian and SQLite indices on btrfs to achieve acceptable performance (even on SSD). Disabling copy-on-write also disables checksumming, thus "raid1" (or higher) configurations may be corrupt after unsafe shutdowns.

Fortunately, these SQLite and Xapian indices are designed to recoverable from git if missing.

Disabling CoW does not prevent all fragmentation. Large values of "publicInbox.indexBatchSize" also limit fragmentation during the initial index.

Avoid snapshotting subvolumes containing Xapian and/or SQLite indices. Snapshots use CoW despite our efforts to disable it, resulting in fragmentation.

filefrag(8) can be used to monitor fragmentation, and "btrfs filesystem defragment -fr $INBOX_DIR" may be necessary.

Large filesystems benefit significantly from the "space_cache=v2" mount option documented in btrfs(5).

Older, non-CoW filesystems are generally work well out-of-the-box for our Xapian and SQLite indices.

Performance on solid state drives

While SSD read performance is generally good, SSD write performance degrades as the drive ages and/or gets full. Issuing "TRIM" commands via fstrim(8) or similar is required to sustain write performance.

Users of the Flash-Friendly File System F2FS <https://en.wikipedia.org/wiki/F2FS> may benefit from optimizations found in SQLite 3.21.0+. Benchmarks are greatly appreciated.

Read-only daemons

public-inbox-httpd(1), public-inbox-imapd(1), and public-inbox-nntpd(1) are all designed for C10K (or higher) levels of concurrency from a single process. SMP systems may use "--worker-processes=NUM" as documented in public-inbox-daemon(8) for parallelism.

The open file descriptor limit ("RLIMIT_NOFILE", "ulimit -n" in sh(1), "LimitNOFILE=" in systemd.exec(5)) may need to be raised to accommodate many concurrent clients.

Transport Layer Security (IMAPS, NNTPS, or via STARTTLS) significantly increases memory use of client sockets, sure to account for that in capacity planning.

Other OS tuning knobs

Linux users: the "sys.vm.max_map_count" sysctl may need to be increased if handling thousands of inboxes (with public-inbox-extindex(1)) to avoid out-of-memory errors from git.

Other OSes may have similar tuning knobs (patches appreciated).

Scalability to many inboxes

public-inbox-extindex(1) allows any number of public-inboxes to share the same Xapian indices.

git 2.33+ startup time is orders-of-magnitude faster and uses less memory when dealing with thousands of alternates required for thousands of inboxes with public-inbox-extindex(1).

Frequent packing (via git-gc(1)) both improves performance and reduces the need to increase "sys.vm.max_map_count".