@node Commands @unnumbered Commands Nearly all commands have the following common options: @table @option @item -cfg Path to configuration file. May be overridden by @env{NNCPCFG} environment variable. If file file is an encrypted @ref{EBlob, eblob}, then ask for passphrase to decrypt it first. @item -debug Print debug messages. Normally this option should not be used. @item -minsize @anchor{OptMinSize} Minimal required resulting packet size, in KiBs. For example if you send 2 KiB file and set @option{-minsize 4}, then resulting packet will be 4 KiB (containing file itself and some junk). @item -nice Set desired outgoing packet @ref{Niceness, niceness level}. @item -replynice Set desired reply packet @ref{Niceness, niceness level}. Only freq and exec packets look at that niceness level. @item -via Override @ref{CfgVia, via} configuration option for destination node. Specified nodes must be separated with comma: @verb{|NODE1,NODE2|}. With @verb{|-via -|} you can disable relaying at all. @item -spool Override path to spool directory. May be specified by @env{NNCPSPOOL} environment variable. @item -log Override path to logfile. May be specified by @env{NNCPLOG} environment variable. @item -quiet Print only errors, omit simple informational messages. In any case those messages are logged, so you can reread them using @ref{nncp-log} command. @item -progress, -noprogress Either force progress showing, or disable it. @item -version Print version information. @item -warranty Print warranty information (no warranty). @end table @node nncp-bundle @section nncp-bundle @example $ nncp-bundle [options] -tx [-delete] NODE [NODE ...] > ... $ nncp-bundle [options] -rx -delete [-dryrun] [NODE ...] < ... $ nncp-bundle [options] -rx [-check] [-dryrun] [NODE ...] < ... @end example With @option{-tx} option, this command creates @ref{Bundles, bundle} of @ref{Encrypted, encrypted packets} from the spool directory and writes it to @code{stdout}. With @option{-rx} option, this command takes bundle from @code{stdin} and copies all found packets for our node to the spool directory. Pay attention that @strong{no} integrity checking is done by default. Modern tape drives could easily provide too much throughput your CPU won't be able to verify on the fly. So if you won't @ref{nncp-toss, toss} received packets at the place, it is advisable either to run @ref{nncp-check} utility for packets integrity verification, or to use @option{-check} option to enable on the fly integrity check. You can specify multiple @option{NODE} arguments, telling for what nodes you want to create the stream, or take it from. If no nodes are specified for @option{-rx} mode, then all packets aimed at us will be processed. When packets are sent through the stream, they are still kept in the spool directory, because there is no assurance that they are transferred to the media (media (CD-ROM, tape drive, raw hard drive) can end). If you want to forcefully delete them (after they are successfully flushed to @code{stdout}) anyway, use @option{-delete} option. But you can verify produced stream after, by digesting it by yourself with @option{-rx} and @option{-delete} options -- in that mode, stream packets integrity will be checked and they will be deleted from the spool if everything is good. So it is advisable to recheck your streams: @example $ nncp-bundle -tx ALICE BOB WHATEVER | cdrecord -tao - $ dd if=/dev/cd0 bs=2048 | nncp-bundle -rx -delete @end example @option{-dryrun} option prevents any writes to the spool. This is useful when you need to see what packets will pass by and possibly check their integrity. @node nncp-call @section nncp-call @example $ nncp-call [options] [-onlinedeadline INT] [-maxonlinetime INT] [-rx|-tx] [-list] [-pkts PKT,PKT,...] [-rxrate INT] [-txrate INT] [-autotoss*] NODE[:ADDR] [FORCEADDR] @end example Call (connect to) specified @option{NODE} and run @ref{Sync, synchronization} protocol with the @ref{nncp-daemon, daemon} on the remote side. Normally this command could be run any time you wish to either check for incoming packets, or to send out queued ones. Synchronization protocol allows resuming and bidirectional packets transfer. If @option{-rx} option is specified then only inbound packets transmission is performed. If @option{-tx} option is specified, then only outbound transmission is performed. @option{-onlinedeadline} overrides @ref{CfgOnlineDeadline, @emph{onlinedeadline}}. @option{-maxonlinetime} overrides @ref{CfgMaxOnlineTime, @emph{maxonlinetime}}. @option{-rxrate}/@option{-txrate} override @ref{CfgXxRate, rxrate/txrate}. @option{-list} option allows you to list packets of remote node, without any transmission. You can specify what packets your want to download, by specifying @option{-pkts} option with comma-separated list of packets identifiers. Each @option{NODE} can contain several uniquely identified @option{ADDR}esses in @ref{CfgAddrs, configuration} file. If you do not specify the exact one, then all will be tried until the first success. Optionally you can force @option{FORCEADDR} address usage, instead of addresses taken from configuration file. You can specify both @verb{|host:port|} and @verb{#|some command#} formats. Pay attention that this command runs integrity check for each completely received packet in the background. This can be time consuming. Connection could be lost during that check and remote node won't be notified that file is done. But after successful integrity check that file is renamed from @file{.part} one and when you rerun @command{nncp-call} again, remote node will receive completion notification. @option{-autotoss} options runs tosser on node's spool after call is finished. All @option{-autotoss-*} options is the same as in @ref{nncp-toss} command. @node nncp-caller @section nncp-caller @example $ nncp-caller [options] [NODE ...] @end example Croned daemon that calls remote nodes from time to time, according to their @ref{CfgCalls, @emph{calls}} configuration field. Optional number of @option{NODE}s tells to ignore other ones. Otherwise all nodes with specified @emph{calls} configuration field will be called. Look at @ref{nncp-call} for more information. @node nncp-cfgenc @section nncp-cfgenc @example $ nncp-cfgenc [options] [-s INT] [-t INT] [-p INT] cfg.hjson > cfg.hjson.eblob $ nncp-cfgenc [options] -d cfg.hjson.eblob > cfg.hjson @end example This command allows you to encrypt provided @file{cfg.hjson} file with the passphrase, producing @ref{EBlob, eblob}, to safely keep your configuration file with private keys. This utility was written for users who do not want (or can not) to use either @url{https://gnupg.org/, GnuPG} or similar tools. That @file{eblob} file can be used directly in @option{-cfg} option of nearly all commands. @option{-s}, @option{-t}, @option{-p} are used to tune @file{eblob}'s password strengthening function. Space memory cost (@option{-s}), specified in number of BLAKE2b-256 blocks (32 bytes), tells how many memory must be used for hashing -- bigger values are better, but slower. Time cost (@option{-t}) tells how many rounds/iterations must be performed -- bigger is better, but slower. Number of parallel jobs (@option{-p}) tells how many computation processes will be run: this is the same as running that number of independent hashers and then joining their result together. When invoked for encryption, passphrase is entered manually twice. When invoked for decryption (@option{-d} option), it is asked once and exits if passphrase can not decrypt @file{eblob}. @option{-dump} options parses @file{eblob} and prints parameters used during its creation. For example: @example $ nncp-cfgenc -dump /usr/local/etc/nncp.hjson.eblob Strengthening function: Balloon with BLAKE2b-256 Memory space cost: 1048576 bytes Number of rounds: 16 Number of parallel jobs: 2 Blob size: 2494 @end example @node nncp-cfgmin @section nncp-cfgmin @example $ nncp-cfgmin [options] > stripped.hjson @end example Print out stripped configuration version: only path to @ref{Spool, spool}, path to log file, neighbours public keys are stayed. This is useful mainly for usage with @ref{nncp-xfer} that has to know only neighbours, without private keys involving. @node nncp-cfgnew @section nncp-cfgnew @example $ nncp-cfgnew [options] [-nocomments] > new.hjson @end example Generate new node configuration: private keys, example configuration file and print it to @code{stdout}. You must use this command when you setup the new node. @option{-nocomments} will create configuration file without descriptive huge comments -- useful for advanced users. Pay attention that private keys generation consumes an entropy from your operating system. @node nncp-check @section nncp-check @example $ nncp-check [options] @end example Perform @ref{Spool, spool} directory integrity check. Read all files that has Base32-encoded filenames and compare it with recalculated BLAKE2b hash output of their contents. That supplementary command is not used often in practice, if ever. @node nncp-daemon @section nncp-daemon @example $ nncp-daemon [options] [-maxconn INT] [-bind ADDR] [-inetd] [-autotoss*] @end example Start listening TCP daemon, wait for incoming connections and run @ref{Sync, synchronization protocol} with each of them. You can run @ref{nncp-toss} utility in background to process inbound packets from time to time. @option{-maxconn} option specifies how many simultaneous clients daemon can handle. @option{-bind} option specifies @option{addr:port} it must bind to and listen. It could be run as @command{inetd} service, by specifying @option{-inetd} option. Pay attention that because it uses @code{stdin}/@code{stdout}, it can not effectively work with IO timeouts and connection closing can propagate up to 5 minutes in practice. Example inetd-entry: @verbatim uucp stream tcp6 nowait nncpuser /usr/local/bin/nncp-daemon nncp-daemon -quiet -inetd @end verbatim @option{-autotoss} options runs tosser on node's spool after call is finished. All @option{-autotoss-*} options is the same as in @ref{nncp-toss} command. @node nncp-exec @section nncp-exec @example $ nncp-exec [options] [-use-tmp] [-nocompress] NODE HANDLE [ARG0 ARG1 ...] @end example Send execution command to @option{NODE} for specified @option{HANDLE}. Body is read from @code{stdin} into memory and compressed (unless @option{-nocompress} is specified). After receiving, remote side will execute specified @ref{CfgExec, handle} command with @option{ARG*} appended and decompressed body fed to command's @code{stdin}. If @option{-use-tmp} option is specified, then @code{stdin} data is read into temporary file first, requiring twice more disk space, but no memory requirements. @ref{StdinTmpFile, Same temporary file} rules applies as with @ref{nncp-file -} command. For example, if remote side has following configuration file for your node: @verbatim exec: { sendmail: [/usr/sbin/sendmail, "-t"] appender: ["/bin/sh", "-c", "cat >> /append"] } @end verbatim then executing @verb{|echo My message | nncp-exec -replynice 123 REMOTE sendmail root@localhost|} will lead to execution of: @example echo My message | NNCP_SELF=REMOTE \ NNCP_SENDER=OurNodeId \ NNCP_NICE=123 \ /usr/sbin/sendmail -t root@@localhost @end example If @ref{CfgNotify, notification} is enabled on the remote side for exec handles, then it will sent simple letter after successful command execution with its output in message body. @strong{Pay attention} that packet generated with this command won't be be chunked. @node nncp-file @section nncp-file @example $ nncp-file [options] [-chunked INT] SRC NODE:[DST] @end example Send @file{SRC} file to remote @option{NODE}. @file{DST} specifies destination file name in remote's @ref{CfgIncoming, incoming} directory. If this file already exists there, then counter will be appended to it. This command queues file in @ref{Spool, spool} directory immediately (through the temporary file of course) -- so pay attention that sending 2 GiB file will create 2 GiB outbound encrypted packet. @anchor{StdinTmpFile} If @file{SRC} equals to @file{-}, then create an encrypted temporary file and copy everything taken from @code{stdin} to it and use for outbound packet creation. Pay attention that if you want to send 1 GiB of data taken from @code{stdin}, then you have to have more than 2 GiB of disk space for that temporary file and resulting encrypted packet. You can control temporary file location directory with @env{TMPDIR} environment variable. Encryption is performed in AEAD mode with @url{https://cr.yp.to/chacha.html, ChaCha20}-@url{https://en.wikipedia.org/wiki/Poly1305, Poly1305} algorithms. Data is splitted on 128 KiB blocks. Each block is encrypted with increasing nonce counter. File is deletes immediately after creation, so even if program crashes -- disk space will be reclaimed, no need in cleaning it up later. If @file{SRC} points to directory, then @url{https://pubs.opengroup.org/onlinepubs/9699919799/utilities/pax.html#tag_20_92_13_01, pax archive} will be created on the fly with directory contents and destination filename @file{.tar} appended. It @strong{won't} contain any entities metainformation, but modification time with the names. UID/GID are set to zero. Directories have 777 permissions, files have 666, for being friendly with @command{umask}. Also each entity will have comment like @verb{|Autogenerated by NNCP version X.Y.Z built with goXXX|}. If @option{-chunked} is specified, then source file will be split @ref{Chunked, on chunks}. @option{INT} is the desired chunk size in KiBs. This mode is more CPU hungry. Pay attention that chunk is saved in spool directory immediately and it is not deleted if any error occurs. @option{-minsize} option is applied per each chunk. Do not forget about @ref{ChunkedZFS, possible} ZFS deduplication issues. Zero @option{-chunked} disables chunked transmission. If @ref{CfgNotify, notification} is enabled on the remote side for file transmissions, then it will sent simple letter after successful file receiving. @node nncp-freq @section nncp-freq @example $ nncp-freq [options] NODE:SRC [DST] @end example Send file request to @option{NODE}, asking it to send its @file{SRC} file from @ref{CfgFreq, freq.path} directory to our node under @file{DST} filename in our @ref{CfgIncoming, incoming} one. If @file{DST} is not specified, then last element of @file{SRC} will be used. If @ref{CfgNotify, notification} is enabled on the remote side for file request, then it will sent simple letter after successful file queuing. @node nncp-log @section nncp-log @example $ nncp-log [options] @end example Parse @ref{Log, log} file and print out its records in human-readable form. @node nncp-pkt @section nncp-pkt @example $ nncp-pkt [options] < pkt $ nncp-pkt [options] [-decompress] -dump < pkt > payload $ nncp-pkt -overheads @end example Low level packet parser. Normally it should not be used, but can help in debugging. By default it will print packet's type, for example: @example Packet type: encrypted Niceness: 64 Sender: 2WHBV3TPZHDOZGUJEH563ZEK7M33J4UESRFO4PDKWD5KZNPROABQ @end example If you specify @option{-dump} option and provide an @ref{Encrypted, encrypted} packet, then it will verify and decrypt it to @code{stdout}. Encrypted packets contain @ref{Plain, plain} ones, that also can be fed to @command{nncp-pkt}: @example Packet type: plain Payload type: transitional Path: VHMTRWDOXPLK7BR55ICZ5N32ZJUMRKZEMFNGGCEAXV66GG43PEBQ Packet type: plain Payload type: mail Path: stargrave@@stargrave.org @end example And with the @option{-dump} option it will give you the actual payload (the whole file, mail message, and so on). @option{-decompress} option tries to zstd-decompress the data from plain packet (useful for mail packets). @option{-overheads} options print encrypted, plain and size header overheads. @node nncp-reass @section nncp-reass @example $ nncp-reass [options] [-dryrun] [-keep] [-dump] [-stdout] FILE.nncp.meta $ nncp-reass [options] [-dryrun] [-keep] @{-all | -node NODE@} @end example Reassemble @ref{Chunked, chunked file} after @ref{nncp-toss, tossing}. When called with @option{FILE} option, this command will reassemble only it. When called with @option{-node} option, this command will try to reassemble all @file{.nncp.meta} files found in @option{NODE}'s @ref{CfgIncoming, incoming} directory. When called with @option{-all} option, then cycle through all known nodes to do the same. Reassembling process does the following: @enumerate @item Parses @ref{Chunked, @file{.nncp.meta}} file. @item Checks existence and size of every @file{.nncp.chunkXXX}. @item Verifies integrity of every chunk. @item Concatenates all chunks, simultaneously removing them from filesystem. @end enumerate That process reads the whole data twice. Be sure to have free disk space for at least one chunk. Decrypted chunk files as a rule are saved in pseudo-random order, so removing them during reassembly process will likely lead to filesystem fragmentation. Reassembly process on filesystems with deduplication capability should be rather lightweight. If @option{-dryrun} option is specified, then only existence and integrity checking are performed. If @option{-keep} option is specified, then no @file{.nncp.meta}/@file{.nncp.chunkXXX} files are deleted during reassembly process. @option{-stdout} option outputs reassembled file to @code{stdout}, instead of saving to temporary file with renaming after. This could be useful for reassembling on separate filesystem to lower fragmentation effect, and/or separate storage device for higher performance. @option{-dump} option prints meta-file contents in human-friendly form. It is useful mainly for debugging purposes. For example: @example Original filename: testfile File size: 3.8 MiB (3987795 bytes) Chunk size: 1.0 MiB (1048576 bytes) Number of chunks: 4 Checksums: 0: eac60d819edf40b8ecdacd0b9a5a8c62de2d15eef3c8ca719eafa0be9b894017 1: 013a07e659f2e353d0e4339c3375c96c7fffaa2fa00875635f440bbc4631052a 2: f4f883975a663f2252328707a30e71b2678f933b2f3103db8475b03293e4316e 3: 0e9e229501bf0ca42d4aa07393d19406d40b179f3922a3986ef12b41019b45a3 @end example Do not forget about @ref{ChunkedZFS, possible} ZFS deduplication issues. @node nncp-rm @section nncp-rm @example $ nncp-rm [options] -tmp $ nncp-rm [options] -lock $ nncp-rm [options] -node NODE -part $ nncp-rm [options] -node NODE -seen $ nncp-rm [options] -node NODE [-rx] [-tx] $ nncp-rm [options] -node NODE -pkt PKT @end example This command is aimed to delete various files from your spool directory: @itemize @item If @option{-tmp} option is specified, then it will delete all temporary files in @file{spool/tmp} directory. Files may stay in it when commands like @ref{nncp-file} fail for some reason. @item If @option{-lock} option is specified, then all @file{.lock} files will be deleted in your spool directory. @item If @option{-pkt} option is specified, then @file{PKT} packet (its Base32 name) will be deleted. This is useful when you see some packet failing to be processed. @item When either @option{-rx} or @option{-tx} options are specified (maybe both of them), then delete all packets from that given queues. If @option{-part} is given, then delete only @file{.part}ly downloaded ones. If @option{-seen} option is specified, then delete only @file{.seen} files. @item @option{-dryrun} option just prints what will be deleted. @item You can also select files that only have modification date older than specified @option{-older} time units (@code{10s} (10 seconds), @code{5m} (5 minutes), @code{12h} (12 hours), @code{2d} (2 days)). @end itemize @node nncp-stat @section nncp-stat @example $ nncp-stat [options] [-pkt] [-node NODE] @end example Print current @ref{Spool, spool} statistics about unsent and unprocessed packets. For each node (unless @option{-node} specified) and each niceness level there will be printed how many packets (with the total size) are in inbound (Rx) and outbound (Tx) queues. @option{-pkt} option show information about each packet. @node nncp-toss @section nncp-toss @example $ nncp-toss [options] [-node NODE] [-dryrun] [-cycle INT] [-seen] [-nofile] [-nofreq] [-noexec] [-notrns] @end example Perform "tossing" operation on all inbound packets. This is the tool that decrypts all packets and processes all payload packets in them: copies files, sends mails, sends out file requests and relays transition packets. It should be run after each online/offline exchange. @option{-dryrun} option does not perform any writing and sending, just tells what it will do. @option{-cycle} option tells not to quit, but to repeat tossing every @option{INT} seconds in an infinite loop. That can be useful when running this command as a daemon. @option{-seen} option creates empty @file{XXX.seen} file after successful tossing of @file{XXX} packet. @ref{nncp-xfer}, @ref{nncp-bundle}, @ref{nncp-daemon} and @ref{nncp-call} commands skip inbound packets that has been already seen, processed and tossed. This is helpful to prevent duplicates. @option{-nofile}, @option{-nofreq}, @option{-noexec}, @option{-notrns} options allow to disable any kind of packet types processing. @node nncp-xfer @section nncp-xfer @example $ nncp-xfer [options] [-node NODE] [-mkdir] [-keep] [-rx|-tx] DIR @end example Search for directory in @file{DIR} containing inbound packets for us and move them to local @ref{Spool, spool} directory. Also search for known neighbours directories and move locally queued outbound packets to them. This command is used for offline packets transmission. If @option{-mkdir} option is specified, then outbound neighbour(s) directories will be created. This is useful for the first time usage, when storage device does not have any directories tree. If @option{-keep} option is specified, then keep copied files, do not remove them. @option{-rx} option tells only to move inbound packets addressed to us. @option{-tx} option tells exactly the opposite: move only outbound packets. @ref{nncp-cfgmin} could be useful for creating stripped minimalistic configuration file version without any private keys. @file{DIR} directory has the following structure: @file{RECIPIENT/SENDER/PACKET}, where @file{RECIPIENT} is Base32 encoded destination node, @file{SENDER} is Base32 encoded sender node. Also look for @ref{nncp-bundle}, especially if you deal with CD-ROM and tape drives.