diff options
| -rw-r--r-- | driver.c | 35 | ||||
| -rw-r--r-- | fetchmail.h | 1 | ||||
| -rw-r--r-- | fetchmail.man | 4 | ||||
| -rw-r--r-- | options.c | 1 |
4 files changed, 36 insertions, 5 deletions
@@ -141,8 +141,9 @@ static int is_host_alias(const char *name, struct query *ctl) if (outlevel != O_SILENT) putchar('\n'); /* terminate the progress message */ fprintf(stderr, - "fetchmail: nameserver evaporated while looking for `%s' during poll of %s.\n", + "fetchmail: nameserver failure while looking for `%s' during poll of %s.\n", name, ctl->servername); + ctl->errcount++; longjmp(restart, 2); /* try again next poll cycle */ break; } @@ -168,6 +169,10 @@ static int is_host_alias(const char *name, struct query *ctl) case NO_RECOVERY: /* non-recoverable name server error */ case TRY_AGAIN: /* temporary error on authoritative server */ default: + fprintf(stderr, + "fetchmail: nameserver failure while looking for `%s' during poll of %s.\n", + name, ctl->servername); + ctl->errcount++; longjmp(restart, 2); /* try again next poll cycle */ break; } @@ -778,6 +783,7 @@ const struct method *proto; /* protocol method table */ return(PS_ERROR); } + if (check_only) { if (new == -1 || ctl->fetchall) @@ -787,12 +793,35 @@ const struct method *proto; /* protocol method table */ } else if (count > 0) { + /* + * What forces this code is that in POP3 you can't fetch a + * message without having it marked `seen'. + * + * The result is that if there's any kind of transient error + * (DNS lookup failure, or sendmail refusing delivery due to + * process-table limits) the message will be marked "seen" on + * the server without having been delivered. This is not a + * big problem if fetchmail is running in foreground, because + * the user will see a "skipped" message when it next runs and + * get clued in. + * + * But in daemon mode this leads to the message being silently + * ignored forever. This is not acceptable. + * + * We compensate for this by checking the error count from the + * previous pass and forcing all messages to be considered new + * if it's nonzero. + */ + int force_retrieval = (ctl->errcount > 0); + + ctl->errcount = 0; + /* read, forward, and delete messages */ for (num = 1; num <= count; num++) { - int toolarge = msgsizes && msgsizes[num-1]>ctl->limit; + int toolarge = msgsizes && (msgsizes[num-1] > ctl->limit); int fetch_it = ctl->fetchall || - (!(protocol->is_old && (protocol->is_old)(sockfp,ctl,num)) && !toolarge); + (!toolarge && (force_retrieval || !(protocol->is_old && (protocol->is_old)(sockfp,ctl,num))); /* we may want to reject this message if it's old */ if (!fetch_it) diff --git a/fetchmail.h b/fetchmail.h index 93ec9131..b1ebf927 100644 --- a/fetchmail.h +++ b/fetchmail.h @@ -92,6 +92,7 @@ struct query /* internal use */ int active; + int errcount; /* count transient errors in last pass */ struct query *next; /* next query control block in chain */ struct query *leader; /* pointer to this query's SMTP leader */ FILE *smtp_sockfp; /* socket descriptor for SMTP connection */ diff --git a/fetchmail.man b/fetchmail.man index 054225ef..be3c558c 100644 --- a/fetchmail.man +++ b/fetchmail.man @@ -332,7 +332,9 @@ The or --nodetach option suppresses detachment of the daemon process from its control terminal. This is primarily useful for debugging. .PP -Note that running fetchmail in daemon mode forces the --fetchall option on. +Note that while running in daemon mode, transient errors (such as DNS +failures or sendmail delivery refusals) effectively force the fetchall +option on for the duration of the next polling cycle. This is a robustness feature. It means that if a message is fetched (and thus marked seen by the mailserver) but not delivered locally due to some transient error, it will be re-fetched during the next @@ -130,7 +130,6 @@ struct query *ctl; /* option record to be initialized */ case 'd': case LA_DAEMON: poll_interval = atoi(optarg); - ctl->fetchall = TRUE; break; case 'N': case LA_NODETACH: |