summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
...
* Forgot (ARGH!) to attribute Chris on v3-release too.David Hankins2006-07-091-1/+1
|
* - If the dhclient were to receive a DHCPNAK while it was in the RENEWDavid Hankins2006-07-092-1/+19
| | | | | | | | state (and consequently, had an active, 'bound' address and related configuration options), it would fail to 'tear down' this information before proceeding into INIT state. dhclient now iterates the dhclient- script with the 'EXPIRE' action to cause these teardowns prior to entering INIT state. [ISC-Bugs #16035]
* Pulled down documentation updates made to dhcpd.conf.5 on HEAD -David Hankins2006-07-093-7/+17
| | | | | | ddns-domainname is now more clearly documented. Also, version to '3.0.5-pre', just so we can distinguish for our tests.
* - A bug where leases not in ACTIVE state would get billed to billed classesDavid Hankins2006-06-152-1/+18
| | | | | | (classes with lease limitations) was fixed. Non-active leases OFFERed to clients are no longer billed (but billing is checked before offering). [ISC-Bugs #16168]
* - Some time value size fixes in 3.0.4 brought on from FreeBSD /usr/ports wereDavid Hankins2006-05-173-21/+16
| | | | | | misapplied to server values rather than client values. The server no longer advertises 8-byte lease-time options when on 64-bit platforms. [ISC-Bugs #16036]
* - Added additional fatal error sanity checks surrounding lease bindingDavid Hankins2006-05-052-12/+24
| | | | | state count calculations (free/active counts used for failover pool balancing). [ISC Bugs #15898]
* - The default minimum lease time (if min-lease-time was not specified)David Hankins2006-05-052-1/+5
| | | | | was raised from 0 to 300. 0 is not thought to be sensible, and is known to be damaging. [ISC Bugs #15829]
* - A warning that host statements declared within subnet or shared-networkDavid Hankins2006-05-052-3/+16
| | | | scopes are actually global has been added. [ISC Bugs #15948]
* Release status, version numbers on the extract directory, etc misc forgottenV3-0-4David Hankins2006-04-271-5/+6
| | | | bits.
* Documentation, copyright, and version updates for 3.0.4 final.David Hankins2006-04-277-19/+82
| | | | | | | - 'atsfp' now the first thing people see in the RELNOTES. - 'infamous42md's security patches now identified as such. - remote-port dhcpd.conf directive now documented. - the printf()'d copyrights are now 2004-2006.
* Document Jean Delvare's manpage contribution.David Hankins2006-04-261-0/+5
|
* Manpage patch from user.David Hankins2006-04-261-1/+2
|
* Version to 3.0.4rc1.V3-0-4RC1David Hankins2006-03-093-7/+12
|
* Document the DOCSIS-device-class relay agent suboption per discussion onDavid Hankins2006-03-071-2/+15
| | | | | the dhcp-server mailing list ("the document should describe every option the server can accept").
* dhcp-eval.5 should not SEE ALSO to dhcp-eval.5...it means dhcp-options.David Hankins2006-03-061-2/+2
|
* - More version++ to 3.0.4b3 stuff.V3-0-4-BETA-3David Hankins2006-02-2230-61/+61
| | | | | - Copyrights to 2004-2006 (not 2005) in modules changed since 3.0.4b2, also the LICENSE.
* Version to 3.0.4b3.David Hankins2006-02-221-1/+1
|
* - The host-name option and a few others were moved from "X" format to "t"David Hankins2006-02-162-6/+18
| | | | format to be compatible with new NULL handling functions. [ISC-Bugs #15811]
* Backing out rt15075.David Hankins2006-02-1528-72/+170
|
* - The server was inappropriately sending leases to the RESET state in theDavid Hankins2006-01-062-4/+22
| | | | | | | | | event that multiple active leases were found to match a singly-identified client. This was changed to RELEASED (by accepting a different, ACTIVE binding, the client is implicitly releasing its lease). This repairs a bug wherein secondary servers in failover pairs detecting this condition move leases to RESET, and primaries refuse to accept that state transition (properly). [ISC-Bugs #15718]
* - DHCPINFORM processing is a little more careful about return addressingDavid Hankins2006-01-062-14/+66
| | | | | | | its responses, or if responding via a relay. The INFORM related messages also log the 'effective client ip address' rather than the client's supplied ciaddr (since some clients produce null ciaddrs). [ISC-Bugs #15738]
* - Null-termination sensing for certain clients that unfortunatley requireDavid Hankins2006-01-062-9/+8
| | | | it in DHCPINFORM processing was repaired. [ISC-Bugs #15787]
* Updated dates for 3.0.4b2 release.V3-0-4-BETA-2David Hankins2005-11-032-2/+2
|
* - Changes to trailing NULL removal in 't' option-atoms has been rethought,David Hankins2005-11-012-27/+33
| | | | | | | | it now includes 'd' (domain name) types, and tries hard not to rewind an option beyond the start of the text field it is un-terminating. [ISC-Bugs #15561] - 'T' DHCP atom documented.
* Start moving version to b2.David Hankins2005-10-273-9/+8
|
* - Changes to trailing NULL removal in 't' option-atoms has been rethought,David Hankins2005-10-273-7/+175
| | | | | | it now includes 'd' (domain name) types, and tries hard not to rewind an option beyond the start of the text field it is un-terminating. [ISC-Bugs #15541]
* - Command line parsing in omshell was repaired - it no longer closesDavid Hankins2005-10-273-6/+13
| | | | | | STDIN after reading one line. [ISC-Bugs #15545] - The resolver library no longer closes the /etc/resolv.conf file descriptor it opened twice. [ISC-Bugs #15545]
* Version to 3.0.4, readme relnotes updates to 3.0.4.V3-0-4-BETA-1David Hankins2005-10-143-11/+11
|
* - The server now tries harder to survive the condition where it is unableDavid Hankins2005-10-142-11/+16
| | | | | to open a new lease file to rewrite the lease state database. [ISC-Bugs #15503]
* Properly attribute to Jason Vas Dias and Redhat for rt15293 work.David Hankins2005-10-141-0/+1
|
* - The linux build environment has had some minor improvements - betterDavid Hankins2005-10-142-2/+12
| | | | | | sensing of 64-bit pointer sizes (only used for establishing an icmp_id), and corrections to #if operators regarding LINUX_MAJOR should it ever move to 3.[01].x. [ISC-Bugs #13115]
* - File handlers on configuration state (config files and lease dbs) shouldDavid Hankins2005-10-144-10/+14
| | | | | be treated consistently, regardless of wether TRACING is defined or not. [ISC-Bugs #13676]
* - "append"ing a string onto the end of a "t" type option (such as theDavid Hankins2005-10-144-9/+29
| | | | | | | domain-name field) that had been improperly NULL-terminated by the DHCP server will no longer result in a truncated string containing only the option from the server, and not the expected appended value. [ISC-Bugs #15293]
* - Compilation on older Solaris systems (lacking /usr/include/sys/int_types.h)David Hankins2005-10-142-1/+13
| | | | has been repaired. [ISC-Bugs #15339]
* - The special status of RELEASED/EXPIRED/RESET leases when a serverDavid Hankins2005-10-103-46/+109
| | | | | | | is operating in partner-down was fixed. It no longer requires a lease be twice the MCLT beyond STOS to 'reallocate', and the expiry event to turn these into FREE leases without peer acknowledgement (after STOS+MCLT) has been repaired. [ISC-Bugs #15433]
* - The calculation of the maximum size of DHCP packets no longer includesDavid Hankins2005-10-103-9/+16
| | | | | | | Ethernet framing overhead. The result is that the 'Maximum Message Size' option advertised by clients, or the default value 576, is no longer reduced by 14 bytes, and instead directly reflects the IP level MTU (and the default, minimum allowed IP MTU of 576). [ISC-Bugs #14674]
* - When writing a trace file, the file is now created with permissions 0600,David Hankins2005-10-102-2/+6
| | | | | to help administrators avoid accidentally publicising sensitive config data. [ISC-Bugs #14593]
* - Several minor bugs, largely relating to treating 8-byte time values asDavid Hankins2005-10-109-40/+46
| | | | | | 4-byte entities, have been repaired after careful review of the FreeBSD ports collection's patch set. Thanks to the nameless entities who have contributed to the FreeBSD ports. [ISC-bugs #13722]
* - An OMAPI host/network byte order problem in lease time values has beenDavid Hankins2005-10-072-29/+38
| | | | repaired. [ISC-Bugs #11277]
* Merge rt14960, impossible sanity check, as from the fellow identifyingDavid Hankins2005-10-062-5/+4
| | | | himself as 'infamous42md'.
* Corrections to [ISC-Bugs #15429] after bugfixing and incremental review;David Hankins2005-09-304-10/+17
| | | | atsfp was not being cleared on expired leases.
* - Adjusted lease state validation checks in potential-conflict, toDavid Hankins2005-09-302-8/+20
| | | | | | account for possible clock skew similarly to normal state, and several previously illegal transitions were made legal (ex: active->released). [ISC-Bugs #14834]
* - Failover potential expiry calculations (TSTP) have been corrected. ResultsDavid Hankins2005-09-302-46/+48
| | | | | should be substantially more consistent, and proper given the constraints. [ISC-Bugs #1926]
* - Document STOP, RELEASE, NBI dhclient-script states.David Hankins2005-09-282-3/+19
|
* - Ignore SIT, IRDA, and IEEE1394 interfaces, as they are either nonsensicalDavid Hankins2005-09-283-2/+30
| | | | for DHCP or unsupported (as in the case of 1394). [ISC-Bugs #15464]
* - If the lease db is deleted out from under the daemon, and it moves to rewriteDavid Hankins2005-09-222-5/+14
| | | | | the db, it will go ahead with the operation and move the new db into place once it detects the old db does not exist. [ISC-Bugs #14605]
* - A failover bug that was allowing leases that EXPIRED or were RELEASEDDavid Hankins2005-09-2210-40/+103
| | | | | | | where tsfp and tstp are identical timestamps to languish in these transitional states has been repaired. As a side effect, lease databases should be kept more consistent overall, not just for these transitional states. [ISC-Bugs #15429] [ISC-Support #694]
* Document -e flag.David Hankins2005-09-141-1/+14
|
* Rectify manpage on new next-server default.David Hankins2005-09-121-4/+2
|
* Merging rt15042 after review - __u16 is a kernel typedef, not a userlandDavid Hankins2005-09-082-6/+9
| | | | one. Use u_int16_t instead. Thanks to a patch from Jason Vas Dias.