On Wed, 22 Oct 2008 11:33:58 +0900
FUJITA Tomonori <fujita.tomonori@xxxxxxxxxxxxx> wrote:
> On Wed, 22 Oct 2008 12:50:22 +1100
> "ronnie sahlberg" <ronniesahlberg@xxxxxxxxx> wrote:
>
> > On Wed, Oct 22, 2008 at 12:47 PM, FUJITA Tomonori
> > <fujita.tomonori@xxxxxxxxxxxxx> wrote:
> > > On Tue, 21 Oct 2008 15:56:17 +0200
> > > Tomasz Chmielewski <mangoo@xxxxxxxx> wrote:
> > >
> > >> FUJITA Tomonori schrieb:
> > >>
> > >> (...)
> > >>
> > >> >> # tgtd ; tgtadm --lld iscsi --op new --mode target --tid 1 -T iqn.2008-01.....
> > >> >> tgtadm: can't connect to the tgt daemon, Connection refused
> > >> >> tgtadm: can't send the request to the tgt daemon, Transport endpoint is not connected
> > >> >
> > >> > I guess that 'daemon' program works differently from what you did here
> > >> > (that's what I wrote in the previous mail). But I might be wrong since
> > >> > I don't know how 'daemon' program works (Debian doesn't have the program).
> > >> >
> > >> > For your case, this patch helps (you also need the tgtadm patch that
> > >> > I've just sent)?
> > >>
> > >> It helps here.
> > >>
> > >> # tgtd ; tgtadm --lld iscsi --op new --mode target --tid 1 -T iqn....
> > >> # tgt-admin -s
> > >> Target 1: iqn....
> > >> System information:
> > >> Driver: iscsi
> > >> (...)
> > >>
> > >> Repeated several times, so it seems like it works.
> > >
> > > Yeah, it should work for your case.
> > >
> > >
> > >> Restarting still needs "sleep 1s" ;)
> > >>
> > >> # tgtd ; tgtadm --op delete --mode system ; tgtd
> > >> (null): iscsi_tcp_init(222) unable to bind server socket, Address already in use
> > >> No available low level driver!
> > >> #
> > >
> > > I guess that there is a way to work around this. But I can't recall
> > > now. I need to check a network programing book.
> > >
> >
> > setsockopt()
> > SO_REUSEADDR
>
> Thanks!
>
> I found that I already use SO_REUSEADDR for iSCSI sockets. :)
>
>
> # tgtd ; tgtadm --op delete --mode system ; tgtd
>
>
> I guess, this problem happens in the following way.
>
> 1. execute `tgtadm --op delete --mode system`
> 2. tgtd sets system_active to zero and notifies tgtadm of the
> completion of the request.
> 3. tgtadm returns and the scripts tries to run tgtd again however tgtd
> still runs (it takes some time until tgtd dies after setting
> system_active to zero).
I don't like this patch since it's a bit hacky but it should work.
Please let me know if someone has a better idea.
diff --git a/usr/tgtadm.c b/usr/tgtadm.c
index ce7eb24..836efaa 100644
--- a/usr/tgtadm.c
+++ b/usr/tgtadm.c
@@ -186,15 +186,13 @@ static int ipc_mgmt_connect(int *fd)
strlen(TGT_IPC_NAMESPACE));
err = connect(*fd, (struct sockaddr *) &addr, sizeof(addr));
- if (err < 0) {
- eprintf("can't connect to the tgt daemon, %m\n");
+ if (err < 0)
return errno;
- }
return 0;
}
-static int ipc_mgmt_rsp(int fd)
+static int ipc_mgmt_rsp(int fd, struct tgtadm_req *req)
{
struct tgtadm_rsp rsp;
int err, rest, len;
@@ -223,6 +221,24 @@ retry:
return EINVAL;
}
+ if (req->mode == MODE_SYSTEM && req->op == OP_DELETE) {
+ while (1) {
+ int __fd, ret;
+ struct timeval tv;
+
+ ret = ipc_mgmt_connect(&__fd);
+ if (ret == ECONNREFUSED)
+ break;
+
+ close(__fd);
+
+ tv.tv_sec = 0;
+ tv.tv_usec = 100 * 1000;
+
+ select(0, NULL, NULL, NULL, &tv);
+ }
+ }
+
rest = rsp.len - sizeof(rsp);
if (!rest)
return 0;
@@ -248,8 +264,10 @@ static int ipc_mgmt_req(struct tgtadm_req *req)
int err, fd = 0;
err = ipc_mgmt_connect(&fd);
- if (err < 0)
+ if (err < 0) {
+ eprintf("can't connect to the tgt daemon, %m\n");
goto out;
+ }
err = write(fd, (char *) req, req->len);
if (err < 0) {
@@ -260,7 +278,7 @@ static int ipc_mgmt_req(struct tgtadm_req *req)
dprintf("sent to tgtd %d\n", err);
- err = ipc_mgmt_rsp(fd);
+ err = ipc_mgmt_rsp(fd, req);
out:
if (fd > 0)
close(fd);
--
To unsubscribe from this list: send the line "unsubscribe stgt" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
