gmane.linux.iscsi.open-iscsi

Re: [PATH 0 of 1] Fix check for nice() return value

The Lee-Man — 2013-05-21T21:17:58

Apologies for Subject formatting errors ... I need to wean myself off of OS 
X ...

On Tuesday, May 21, 2013 2:15:00 PM UTC-7, The Lee-Man wrote:

PATCH 1 of 1] correctly check return value of nice()

Lee Duncan — 2013-05-21T21:16:38

The nice() call can return a negative value, since
it returns the previous nice value.

Signed-off-by: Lee Duncan <leeman.duncan-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org>
---
 usr/iscsi_util.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/usr/iscsi_util.c b/usr/iscsi_util.c
index 5e3420e..ac86847 100644
--- a/usr/iscsi_util.c
+++ b/usr/iscsi_util.c
< at >< at > -60,7 +60,8 < at >< at > int oom_adjust(void)
 char path[ISCSI_OOM_PATH_LEN];
 struct stat statb;
 
-if (nice(-10) < 0)
+errno = 0;
+if (nice(-10) == -1 && errno != 0)
 log_debug(1, "Could not increase process priority: %s",
   strerror(errno));

[PATH 0 of 1] Fix check for nice() return value

Lee Duncan — 2013-05-21T21:15:00

I mentioned this before, but I didn't have a good solution at the time.

In usr/iscsi_util.c, nice() is called like this:

if (nice(-10) < 0)
log_debug(...)

The problem is that nice() returns the current nice value,
and that value can legitimately be less than zero, in which
case a spurious log message is generated.

Although I don't like setting errno directly except when in a
library, the nice(2) man page actually suggests this as the
best way to fix the problem.

Therefore, the supplied patch is designed to fix this problem.

--
Ignore the Lee-Man behind the curtain ...

                     "Life's a long song. But the tune ends too soon for us all." -- Ian Anderson

how to optimize open-iscsi over a 802.3ad lacp with four nic (muli-path)(nr_sessions)(MaxConnections)?

Jelle de Jong — 2013-05-16T16:57:35

Hello everybody,

I’m looking to improve my performance of my connections from the kvm quests 
to the iscsi storage. As far as I can see I have to start using multipath 
with node.session.nr_sessions = 4 in open-iscsi, but I want to be sure and 
ask advice first.

I got a LACP bonding setup that is able to have four 1Gb connections going 
to the storage server simultaneously. But I only use one LUN to have one 
big disk that is managed by lvm2 to create lv's for kvm quests and 
snapshots for back-ups.

I tried node.session.nr_sessions = 4 and ended up with four disk devices, 
and since I use lvm2 to manage things, it got mes up with four identical 
disks and duplicated ID's.

If I need to use multipath can somebody be nice enough to share an 
configuration example to optimal performance over four disks.

Below is my are my current settings, please look at them and see if there 
are some other things that are recommended to look at for more performance. 
If I run dd and about 30 quests with IO the performan

how to improve performance when using LACP with 4Gb nics and open-iscsi and posbible multi-path and nr_sessions?

Jelle de Jong — 2013-05-16T10:45:35


Hello everybody,

I’m looking to improve my performance of my connections from the kvm quests to the iscsi storage. As far as I can see I have to start using multipath with node.session.nr_sessions = 4 in open-iscsi, but I want to be sure and ask advice first.

I got a LACP bonding setup that is able to have four 1Gb connections going to the storage server simultaneously. But I only use one LUN to have one big disk that is managed by lvm2 to create lv's for kvm quests and snapshots for back-ups.

I tried node.session.nr_sessions = 4 and ended up with four disk devices, and since I use lvm2 to manage things, it got mes up with four identical disks and duplicated ID's.

If I need to use multipath can somebody be nice enough to share an configuration example to optimal performance over four disks.

Below is my are my current settings, please look at them and see if there are some other things that are recommended to look at for more performance. If I run dd and about 30 quests with IO the performance really

Re: Re: How to contribute changes?

Chris Moore — 2013-05-15T13:07:31

I've wondered that myself.  Fortunately I'm not a lawyer, and ultimately I
don't distribute the product, so I don't have to deal with it. :-)

I do know there are some that do port to another OS.  Since our boards end
up in embedded systems, some of them even roll their own OSes.

I'm also not sure how the whole proprietary driver thing works - for
example, I think nVidia has closed drivers for Linux, but they get away
with it somehow.

I'd sure the lawyers will have a good time figuring it out.  I just make
sure our corporate legal department knows what's going on and leave it up
to them.

Chris


On Tue, May 14, 2013 at 11:52 PM, Ulrich Windl <
Ulrich.Windl-LOEJiUoeGm199bV2EpccEXqWYbMAw+HU< at >public.gmane.org> wrote:

Antw: Re: How to contribute changes?

Ulrich Windl — 2013-05-15T06:52:38

<CADO0a1Qh+jRx0ACV=hbqHhxeZ2+e-R=S47Sm4VrU7vaAevmg=A< at >mail.gmail.com>:

On GPL: I wonder what the customers do if everything else in the Linux kernel is GPL; port the driver to another OS? ;-)

Re: How to contribute changes?

Chris Moore — 2013-05-15T04:32:00

OCS is "One Connect Storage" - Emulex hardware.  The OCS driver is one that
we (Emulex) provide as a reference driver to our customers.

The driver itself is not GPL because some of the customers don't want to be
bound by the GPL requirements if they end up using our source, so we're not
planning to upstream the driver itself.  But it would be handy to have the
user space iscsi tools aware of our driver so the users don't have to patch
the openiscsi source to talk to the driver.

Thanks,

Chris


On Tue, May 14, 2013 at 9:28 PM, Michael Christie <michaelc-hcNo3dDEHLuVc3sceRu5cw< at >public.gmane.org>wrote:

Re: How to contribute changes?

Michael Christie — 2013-05-15T04:28:16

Do you have a kernel driver too? Are you going to get that upstream?

I do not care if it closed source or if you are not going to get it upstream. Just wondering what other iscsi hardware is out there. I will merge the patch either way.

Can you say what ocs stands for?

On May 14, 2013, at 5:48 PM, Chris Moore <cmoore042-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org> wrote:

Re: How to contribute changes?

Chris Moore — 2013-05-14T22:48:57

Sorry for the delayed reply.  Here's the change I'd like to see about
getting in:

Thanks,

Chris


diff --git a/usr/transport.c b/usr/transport.c
index e6e3dfc..83483c6 100644
--- a/usr/transport.c
+++ b/usr/transport.c
< at >< at > -97,6 +97,13 < at >< at > struct iscsi_transport_template qla4xxx = {
  .ep_disconnect = ktransport_ep_disconnect,
 };

+struct iscsi_transport_template ocs = {
+ .name = "ocs",
+ .ep_connect = ktransport_ep_connect,
+ .ep_poll = ktransport_ep_poll,
+ .ep_disconnect = ktransport_ep_disconnect,
+};
+
 static struct iscsi_transport_template *iscsi_transport_templates[] = {
  &iscsi_tcp,
  &iscsi_iser,
< at >< at > -105,6 +112,7 < at >< at > static struct iscsi_transport_template
*iscsi_transport_templates[] = {
  &bnx2i,
  &qla4xxx,
  &be2iscsi,
+ &ocs,
  NULL
 };



On Wed, May 1, 2013 at 8:28 AM, Mike Christie <michaelc-hcNo3dDEHLuVc3sceRu5cw< at >public.gmane.org> wrote:

[PATCH 4/4] ISCSID: Added uip_mgmt_ipc source

Eddie Wai — 2013-05-13T16:42:24

As part of the iscsid to iscsiuio communication socket.


Signed-off-by: Eddie Wai <eddie.wai-dY08KVG/lbpWk0Htik3J/w< at >public.gmane.org>
---
 usr/uip_mgmt_ipc.c |   41 +++++++++++++++++++++++++++++
 usr/uip_mgmt_ipc.h |   73 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 114 insertions(+), 0 deletions(-)
 create mode 100644 usr/uip_mgmt_ipc.c
 create mode 100644 usr/uip_mgmt_ipc.h

diff --git a/usr/uip_mgmt_ipc.c b/usr/uip_mgmt_ipc.c
new file mode 100644
index 0000000..f3074ee
--- /dev/null
+++ b/usr/uip_mgmt_ipc.c
< at >< at > -0,0 +1,41 < at >< at >
+/*
+ * uIP iSCSI Daemon/Admin Management IPC
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published
+ * by the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILIT

Re: [PATCH 1/3] ISCSID: Added socket communication hooks for uip

Mike Christie — 2013-05-13T08:25:29

Getting closer. Could you attach these files? They were missing from the 
patch. Probably just forgot a git add on them when the patch was made.

Re: [PATCH 1/3] iscsiadm: Add flash node mgmt support.

The Lee-Man — 2013-05-09T21:09:37

Looks good to me, and tested as well.

Acked-by: Lee Duncan <leeman.duncan-O5WfVfzUwx8< at >public.gmane.org>

RE: [PATCH 1/3] iscsiadm: Add flash node mgmt support.

Adheer Chandravanshi — 2013-05-09T12:03:44

Mike,

Please find the patch attached herewith that fixes the issue reported by Lee.

Thanks,
Adheer




________________________________

This message and any attached documents contain information from QLogic Corporation or its wholly-owned subsidiaries that may be confidential. If you are not the intended recipient, you may not read, copy, distribute, or use this information. If you have received this transmission in error, please notify the sender immediately by reply e-mail and then delete this message.

[PATCH 6/6] iscsi-tools: Additional parameters for network settings

2013-05-09T10:07:33

From: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>

Signed-off-by: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
---
 include/iscsi_if.h |   75 ++++
 usr/config.h       |   52 +++
 usr/idbm.c         |   91 ++++
 usr/idbm_fields.h  |   53 +++
 usr/iface.c        | 1234 +++++++++++++++++++++++++++++++++++++++++++++++-----
 5 files changed, 1405 insertions(+), 100 deletions(-)

diff --git a/include/iscsi_if.h b/include/iscsi_if.h
index b34a91a..af33b84 100644
--- a/include/iscsi_if.h
+++ b/include/iscsi_if.h
< at >< at > -401,6 +401,81 < at >< at > enum iscsi_net_param {
 ISCSI_NET_PARAM_IFACE_NAME,
 ISCSI_NET_PARAM_MTU,
 ISCSI_NET_PARAM_PORT,
+ISCSI_NET_PARAM_IPADDR_STATE,
+ISCSI_NET_PARAM_IPV6_LINKLOCAL_STATE,
+ISCSI_NET_PARAM_IPV6_ROUTER_STATE,
+ISCSI_NET_PARAM_DELAYED_ACK_EN,
+ISCSI_NET_PARAM_ISNS_EN,
+ISCSI_NET_PARAM_ISNS_ADDR,
+ISCSI_NET_PARAM_ISNS_PORT,
+ISCSI_NET_

[PATCH 3/6] iscsi-tools: Use single function to enable/disable network parameters

2013-05-09T10:07:30

From: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>

Signed-off-by: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
---
 include/iscsi_if.h |  10 ++--
 usr/iface.c        | 143 ++++++++++++++++++++++++++---------------------------
 2 files changed, 72 insertions(+), 81 deletions(-)

diff --git a/include/iscsi_if.h b/include/iscsi_if.h
index 08cfaa9..b34a91a 100644
--- a/include/iscsi_if.h
+++ b/include/iscsi_if.h
< at >< at > -352,9 +352,9 < at >< at > struct iscsi_path {
 uint16_tpmtu;
 } __attribute__ ((aligned (sizeof(uint64_t))));
 
-/* iscsi iface enabled/disabled setting */
-#define ISCSI_IFACE_DISABLE0x01
-#define ISCSI_IFACE_ENABLE0x02
+/* iscsi generic enable/disable setting for various features */
+#define ISCSI_NET_PARAM_DISABLE0x01
+#define ISCSI_NET_PARAM_ENABLE0x02
 
 /* ipv4 bootproto */
 #define ISCSI_BOOTPROTO_STATIC0x01
< at >< at > -379,10 +379,6 < at >< at > struct iscsi_pa

[PATCH 1/6] iscsi-tools: Remove numbers used for network parameter settings

2013-05-09T10:07:28

From: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>

Signed-off-by: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
---
 include/iscsi_if.h | 38 +++++++++++++++++++-------------------
 1 file changed, 19 insertions(+), 19 deletions(-)

diff --git a/include/iscsi_if.h b/include/iscsi_if.h
index 20f2bc2..08cfaa9 100644
--- a/include/iscsi_if.h
+++ b/include/iscsi_if.h
< at >< at > -386,25 +386,25 < at >< at > struct iscsi_path {
 /* iSCSI network params */
 enum iscsi_net_param {
 ISCSI_NET_PARAM_IPV4_ADDR= 1,
-ISCSI_NET_PARAM_IPV4_SUBNET= 2,
-ISCSI_NET_PARAM_IPV4_GW= 3,
-ISCSI_NET_PARAM_IPV4_BOOTPROTO= 4,
-ISCSI_NET_PARAM_MAC= 5,
-ISCSI_NET_PARAM_IPV6_LINKLOCAL= 6,
-ISCSI_NET_PARAM_IPV6_ADDR= 7,
-ISCSI_NET_PARAM_IPV6_ROUTER= 8,
-ISCSI_NET_PARAM_IPV6_ADDR_AUTOCFG= 9,
-ISCSI_NET_PARAM_IPV6_LINKLOCAL_AUTOCFG= 10,
-ISCSI_NET_PARAM_IPV6_ROUTER_AUTOCFG= 11,
-

[PATCH 4/6] iscsi-tools: Use single function to set integer network parameters

2013-05-09T10:07:31

From: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>

Signed-off-by: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
---
 usr/iface.c | 154 ++++++++++++++++++++++++++++++++----------------------------
 1 file changed, 83 insertions(+), 71 deletions(-)

diff --git a/usr/iface.c b/usr/iface.c
index 75ed28e..b77b564 100644
--- a/usr/iface.c
+++ b/usr/iface.c
< at >< at > -1223,60 +1223,68 < at >< at > int iface_get_param_count(struct iface_rec *iface, int iface_all)
 return iface_params.count;
 }
 
-/* IPv4/IPv6 Port: 3260 or User defined */
-static int iface_fill_port(struct iovec *iov, struct iface_rec *iface,
-   uint32_t iface_type)
+/* write integer parameter value */
+static int iface_fill_int_param_val(struct iovec *iov, uint32_t iface_num,
+    uint8_t iface_type, uint16_t param,
+    uint32_t param_len, uint32_t param_val)
 {
 int len;
 struct iscsi_iface_param_inf

[PATCH 2/6] iscsi-tools: Use macro to set IPv4/IPv6 IP addresses

2013-05-09T10:07:29

From: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>

Signed-off-by: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
---
 usr/iface.c | 152 +++++++++++++++++++++++++++---------------------------------
 1 file changed, 67 insertions(+), 85 deletions(-)

diff --git a/usr/iface.c b/usr/iface.c
index c86892e..d3bada4 100644
--- a/usr/iface.c
+++ b/usr/iface.c
< at >< at > -1484,8 +1484,8 < at >< at > static int iface_fill_router_autocfg(struct iovec *iov, struct iface_rec *iface)
 }
 
 /* IPv4 IPAddress/Subnet Mask/Gateway: 4 bytes */
-static int iface_fill_net_ipv4_addr(struct iovec *iov, struct iface_rec *iface,
-    uint32_t param)
+static int iface_fill_net_ipv4_addr(struct iovec *iov, uint32_t iface_num,
+    uint16_t param, char *param_val)
 {
 int rc = 1;
 int len;
< at >< at > -1502,29 +1502,12 < at >< at > static int iface_fill_net_ipv4_addr(struct iovec *iov, struct iface_rec *iface,
 ne

[PATCH 5/6] iscsi-tools: Ignore network parameter if not enabled/disabled

2013-05-09T10:07:32

From: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>

If iface does not contain network parameter or incorrect value is set,
do not send that network parameter value to driver.

Signed-off-by: Harish Zunjarrao <harish.zunjarrao-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
Signed-off-by: Vikas Chaudhary <vikas.chaudhary-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>
---
 usr/iface.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/usr/iface.c b/usr/iface.c
index b77b564..622c436 100644
--- a/usr/iface.c
+++ b/usr/iface.c
< at >< at > -1346,11 +1346,18 < at >< at > static int iface_fill_enable_disable_param(struct iovec *iov,
 net_param->param = param;
 net_param->iface_type = iface_type;
 net_param->param_type = ISCSI_NET_PARAM;
-if (strcmp(param_val, "disable"))
-net_param->value[0] = ISCSI_NET_PARAM_ENABLE;
-else /* Assume disabled */
+if (!strcmp(param_val, "disable"))
 net_param->value[0] = ISCSI_NET_PARAM_DISABLE;
+else if (!strcmp(param_val, "enable"))
+net_

[PATCH 0/6] open-iscsi: Updates for open-iscsi

2013-05-09T10:07:27

From: Vikas Chaudhary <vikas.chaudhary-h88ZbnxC6KDQT0dZR+AlfA< at >public.gmane.org>

Mike,

Please review and apply following patches to open-iscsi.git tree
at your earliest convenience.
These patches include code optimization in iface code as well as
add support to set additional network parameter to iSCSI interface.

Kernel Patches for same is posted here:-
http://marc.info/?l=linux-scsi&m=136809521102233&w=2

Harish Zunjarrao (6):
      iscsi-tools: Remove numbers used for network parameter settings
      iscsi-tools: Use macro to set IPv4/IPv6 IP addresses
      iscsi-tools: Use single function to enable/disable network parameters
      iscsi-tools: Use single function to set integer network parameters
      iscsi-tools: Ignore network parameter if not enabled/disabled
      iscsi-tools: Additional parameters for network settings

Thanks,
Vikas.