DRBD Assistance

[mir]

The problem was 3 functions with normally will not be available in a standard 2.6.32 kernel but since pve uses the 2.6.32 rhel-6.3 kernel these functions has been backported by Redhat.

diff -ru netxtreme2-7.2.20/bnx2x/src/bnx2x_compat.h netxtreme2-7.2.20_new//bnx2x/src/bnx2x_compat.h
--- netxtreme2-7.2.20/bnx2x/src/bnx2x_compat.h    2012-03-28 23:36:48.000000000 +0200
+++ netxtreme2-7.2.20_new//bnx2x/src/bnx2x_compat.h    2012-12-07 18:39:25.969855545 +0100
@@ -824,7 +824,7 @@
 }
 #endif
 
-#if (LINUX_VERSION_CODE < 0x020624)
+/*#if (LINUX_VERSION_CODE < 0x020624)
 static inline void usleep_range(unsigned long min, unsigned long max)
 {
     if (min < 1000)
@@ -832,7 +832,7 @@
     else
         msleep(min / 1000);
 }
-#endif
+#endif*/
 
 #if (LINUX_VERSION_CODE < 0x02061D)
 static inline ssize_t
@@ -1539,7 +1539,7 @@
 }
 #endif
 
-#if (LINUX_VERSION_CODE < KERNEL_VERSION(3, 2, 0))
+/*#if (LINUX_VERSION_CODE < KERNEL_VERSION(3, 2, 0))
 static inline struct page *skb_frag_page(const skb_frag_t *frag)
 {
     return frag->page;
@@ -1553,7 +1553,7 @@
     return dma_map_page(dev, skb_frag_page(frag),
                 frag->page_offset + offset, size, dir);
 }
-#endif
+#endif*/
 
diff -ru netxtreme2-7.2.20/bnx2x-1.72.18/src/bnx2x_compat.h netxtreme2-7.2.20_new//bnx2x-1.72.18/src/bnx2x_compat.h
--- netxtreme2-7.2.20/bnx2x-1.72.18/src/bnx2x_compat.h    2012-03-28 23:36:48.000000000 +0200
+++ netxtreme2-7.2.20_new//bnx2x-1.72.18/src/bnx2x_compat.h    2012-12-07 18:39:25.969855545 +0100
@@ -824,7 +824,7 @@
 }
 #endif
 
-#if (LINUX_VERSION_CODE < 0x020624)
+/*#if (LINUX_VERSION_CODE < 0x020624)
 static inline void usleep_range(unsigned long min, unsigned long max)
 {
     if (min < 1000)
@@ -832,7 +832,7 @@
     else
         msleep(min / 1000);
 }
-#endif
+#endif*/
 
 #if (LINUX_VERSION_CODE < 0x02061D)
 static inline ssize_t
@@ -1539,7 +1539,7 @@
 }
 #endif
 
-#if (LINUX_VERSION_CODE < KERNEL_VERSION(3, 2, 0))
+/*#if (LINUX_VERSION_CODE < KERNEL_VERSION(3, 2, 0))
 static inline struct page *skb_frag_page(const skb_frag_t *frag)
 {
     return frag->page;
@@ -1553,7 +1553,7 @@
     return dma_map_page(dev, skb_frag_page(frag),
                 frag->page_offset + offset, size, dir);
 }
-#endif
+#endif*/
 
 #if (LINUX_VERSION_CODE >= KERNEL_VERSION(3, 0, 0)) || \
     (defined(RHEL_RELEASE_CODE) && RHEL_RELEASE_CODE >= RHEL_RELEASE_VERSION(6, 2))

 

[adamb]

The new driver is working gggreaat!! Everything is falling into place now!

root@medprox1:~# ethtool -i eth0
driver: bnx2x
version: 1.72.18
firmware-version: bc 6.2.15 phy 4.f
bus-info: 0000:20:00.0

 

[adamb]

Just thinking about this some. Until the latest driver is in the Kernel, I will need to be careful when updating these nodes correct? Potentially the driver I complied could get over written by an update? I appreciate the input!

 

[mir]

Keep the source folder a safe place and remember after each kernel upgrade to compile and install the driver. Btw did you remember to install the driver into the boot initrd?
If not the procedure is as follows:
sudo update-initramfs -u

This also needs to be done after every kernel upgrade.

 

[mo_]

have you opened a bug report about this and posted your driver update as a proposed fix yet? ... to get it added to the main repository.
also I will test them out too for the time being (the blade only serves as a test machine anyway)

 

[adamb]

have you opened a bug report about this and posted your driver update as a proposed fix yet? ... to get it added to the main repository.
also I will test them out too for the time being (the blade only serves as a test machine anyway)

No I was told they use the drivers in the rhel kernel, so that would not do any good.

 

[mo_]

well. I wouldnt consider leaving faulty drivers that renders proxmox on blades completely unusable a viable solution.

However I think it was said that newer drivers appeared to have issues? What were those issues? Im willing to specifically test for them (if possible)

 

[tom]

well. I wouldnt consider leaving faulty drivers that renders proxmox on blades completely unusable a viable solution.

However I think it was said that newer drivers appeared to have issues? What were those issues? Im willing to specifically test for them (if possible)

Subject of this thread is DRBD, so if topic changes, pls open a new thread.

 

[adamb]

well. I wouldnt consider leaving faulty drivers that renders proxmox on blades completely unusable a viable solution.

However I think it was said that newer drivers appeared to have issues? What were those issues? Im willing to specifically test for them (if possible)

I am using the newwest drivers without issues. Its the ones in the proxmox kernel which are the issue. I believe they are 1.72.0